diff --git a/index.html b/index.html
index 1e19888..e2909ab 100644
--- a/index.html
+++ b/index.html
@@ -727,6 +727,7 @@ Loading transcript
+
diff --git a/javascript/display.js b/javascript/display.js
index c7e203c..b9064ff 100644
--- a/javascript/display.js
+++ b/javascript/display.js
@@ -43,35 +43,35 @@
// format seconds to minutes
function fmtMSS(s) {
- return (s - (s %= 60)) / 60 + (9 < s ? ':' : ':0') + Math.round(s)
+ return (s - (s %= 60)) / 60 + (9 < s ? ':' : ':0') + Math.round(s)
}
// clear transcript text
// TODO 1. add warning before clearing; 2. make it possible to undo
function clearTranscript() {
- $('#content').html('')
- $('.annotation-content').html('');
+ $('#content').html('')
+ $('.annotation-content').html('');
}
// create a new paragraph tag
function CreateNewPara(timeOfFirstWord, speaker, paraId) {
- var formattedTime = fmtMSS(timeOfFirstWord)
- var paraTime = "";
- // only give it span if it's a word?
- var paraSpeaker = "" + speaker + " ";
- var paraFormattedTime = "[" + formattedTime + "] ";
- var endPara = "
"
- var newPara = paraTime + paraSpeaker + paraFormattedTime + endPara;
- return newPara;
+ var formattedTime = fmtMSS(timeOfFirstWord)
+ var paraTime = "";
+ // only give it span if it's a word?
+ var paraSpeaker = "" + speaker + " ";
+ var paraFormattedTime = "[" + formattedTime + "] ";
+ var endPara = "
"
+ var newPara = paraTime + paraSpeaker + paraFormattedTime + endPara;
+ return newPara;
}
// load audio from file or url using the dropdown or text input
function getAudioUrl() {
- // from file in data folder
- // var audioUrl = "data/" + document.getElementById("audio-name").value;
- var audioUrl = document.getElementById("audioUrll").value;
- document.getElementById("hyperplayer").src = audioUrl;
+ // from file in data folder
+ // var audioUrl = "data/" + document.getElementById("audio-name").value;
+ var audioUrl = document.getElementById("audioUrll").value;
+ document.getElementById("hyperplayer").src = audioUrl;
}
// load audio from user selected file
@@ -79,50 +79,50 @@ function getAudioUrl() {
// set limit
function handleFileSelect(evt) {
- var files = evt.target.files; // FileList object
-
-
-
- // Loop through the FileList and render image files as thumbnails.
- for (var i = 0, f; f = files[i]; i++) {
-
- // Only process audio files.
- if (!f.type.match('audio.*')) {
- alert('Must be an audio file');
- return;
- };
-
- // Warn user if file size is larger than 50mb
- fileSize = Math.round(files[0].size / 1000000)
- if (fileSize > 50.457280) {
- alert('This audio file is ' + fileSize + 'mb. There may be playback issues above 50mb. For ways to reduce file size see the docs: https://github.com/smlum/scription#uploading-large-audio-files-to-scription');
- // TODO use Bulma modal instead of alert
- // $('.modal-audio-size'){}
- // return;
- }
-
- var reader = new FileReader();
-
- // Closure to capture the file information.
- reader.onload = (function (theFile) {
- return function (e) {
- // Render thumbnail.
-
- document.getElementById("hyperplayer").src = e.target.result;
- console.log("loading audio: " + theFile.name);
- document.getElementById("audio-name").innerHTML = theFile.name;
-
- // var span = document.createElement('span');
- // span.innerHTML = ['
'
- // ].join('');
- // document.getElementById('list').insertBefore(span, null);
- };
- })(f);
-
- // Read in the image file as a data URL.
- reader.readAsDataURL(f);
- }
+ var files = evt.target.files; // FileList object
+
+
+
+ // Loop through the FileList and render image files as thumbnails.
+ for (var i = 0, f; f = files[i]; i++) {
+
+ // Only process audio files.
+ if (!f.type.match('audio.*')) {
+ alert('Must be an audio file');
+ return;
+ };
+
+ // Warn user if file size is larger than 50mb
+ fileSize = Math.round(files[0].size / 1000000)
+ if (fileSize > 50.457280) {
+ alert('This audio file is ' + fileSize + 'mb. There may be playback issues above 50mb. For ways to reduce file size see the docs: https://github.com/smlum/scription#uploading-large-audio-files-to-scription');
+ // TODO use Bulma modal instead of alert
+ // $('.modal-audio-size'){}
+ // return;
+ }
+
+ var reader = new FileReader();
+
+ // Closure to capture the file information.
+ reader.onload = (function (theFile) {
+ return function (e) {
+ // Render thumbnail.
+
+ document.getElementById("hyperplayer").src = e.target.result;
+ console.log("loading audio: " + theFile.name);
+ document.getElementById("audio-name").innerHTML = theFile.name;
+
+ // var span = document.createElement('span');
+ // span.innerHTML = ['
'
+ // ].join('');
+ // document.getElementById('list').insertBefore(span, null);
+ };
+ })(f);
+
+ // Read in the image file as a data URL.
+ reader.readAsDataURL(f);
+ }
}
// check for user file entry
@@ -151,106 +151,106 @@ document.getElementById('user-audio-file').addEventListener('change', handleFile
// load json from user selected file
function handleJsonFileSelect(evt) {
- var files = evt.target.files; // FileList object
-
- // console.log(files);
-
- // Loop through the FileList and render image files as thumbnails.
- for (var i = 0, f; f = files[i]; i++) {
-
- // Only process audio files.
- // if (!f.type.match('json.*')) {
- // console.log('json detected');
- // continue;
- // }
-
- var reader = new FileReader();
-
- // console.log(f.name);
- // document.getElementById("json-name").innerHTML = f.name;
- // console.log(f.target.result);
- // displayTranscript();
-
- // Closure to capture the file information.
- reader.onload = (function (theFile) {
- return function (e) {
- // Render thumbnail.
-
- console.log("loading json: " + theFile.name);
- document.getElementById("json-name").innerHTML = theFile.name;
- // console.log(e);
- JsonObj = JSON.parse(e.target.result);
- // console.log(JsonObj);
- displayTranscript(JsonObj);
-
- // var span = document.createElement('span');
- // span.innerHTML = [''
- // ].join('');
- // document.getElementById('list').insertBefore(span, null);
- };
- })(f);
-
- // Read in the image file as a data URL.
- reader.readAsText(f);
- }
-}
+ var files = evt.target.files; // FileList object
-// Listen for user uploading a json
-document.getElementById('user-json-file').addEventListener('change', handleJsonFileSelect, false);
+ // console.log(files);
+ // Loop through the FileList and render image files as thumbnails.
+ for (var i = 0, f; f = files[i]; i++) {
-// load project from user selected file
-function handleProjectFileSelect(evt) {
- var files = evt.target.files; // FileList object
+ // Only process audio files.
+ // if (!f.type.match('json.*')) {
+ // console.log('json detected');
+ // continue;
+ // }
- console.log(files);
+ var reader = new FileReader();
- // Loop through the FileList and render image files as thumbnails.
- for (var i = 0, f; f = files[i]; i++) {
+ // console.log(f.name);
+ // document.getElementById("json-name").innerHTML = f.name;
+ // console.log(f.target.result);
+ // displayTranscript();
- // Only process audio files.
- // if (!f.type.match('json.*')) {
- // console.log('json detected');
- // continue;
- // }
+ // Closure to capture the file information.
+ reader.onload = (function (theFile) {
+ return function (e) {
+ // Render thumbnail.
- var reader = new FileReader();
-
- // console.log(f.name);
- // document.getElementById("json-name").innerHTML = f.name;
- // console.log(f.target.result);
- // displayTranscript();
-
- // Closure to capture the file information.
- reader.onload = (function (theFile) {
- return function (e) {
- // Render thumbnail.
+ console.log("loading json: " + theFile.name);
+ document.getElementById("json-name").innerHTML = theFile.name;
+ // console.log(e);
+ JsonObj = JSON.parse(e.target.result);
+ // console.log(JsonObj);
+ displayTranscript(JsonObj);
+
+ // var span = document.createElement('span');
+ // span.innerHTML = [''
+ // ].join('');
+ // document.getElementById('list').insertBefore(span, null);
+ };
+ })(f);
+
+ // Read in the image file as a data URL.
+ reader.readAsText(f);
+ }
+}
- console.log("loading project:" + theFile.name);
- document.getElementById("project-name").innerHTML = theFile.name;
- $("#content").html(e.target.result)
+// Listen for user uploading a json
+document.getElementById('user-json-file').addEventListener('change', handleJsonFileSelect, false);
- // var span = document.createElement('span');
- // span.innerHTML = [''
- // ].join('');
- // document.getElementById('list').insertBefore(span, null);
- };
- })(f);
- // Read in the image file as a data URL.
- reader.readAsText(f);
- }
+// load project from user selected file
+function handleProjectFileSelect(evt) {
+ var files = evt.target.files; // FileList object
+
+ console.log(files);
+
+ // Loop through the FileList and render image files as thumbnails.
+ for (var i = 0, f; f = files[i]; i++) {
+
+ // Only process audio files.
+ // if (!f.type.match('json.*')) {
+ // console.log('json detected');
+ // continue;
+ // }
+
+ var reader = new FileReader();
+
+ // console.log(f.name);
+ // document.getElementById("json-name").innerHTML = f.name;
+ // console.log(f.target.result);
+ // displayTranscript();
+
+ // Closure to capture the file information.
+ reader.onload = (function (theFile) {
+ return function (e) {
+ // Render thumbnail.
+
+ console.log("loading project:" + theFile.name);
+ document.getElementById("project-name").innerHTML = theFile.name;
+ $("#content").html(e.target.result)
+
+ // var span = document.createElement('span');
+ // span.innerHTML = [''
+ // ].join('');
+ // document.getElementById('list').insertBefore(span, null);
+ };
+ })(f);
+
+ // Read in the image file as a data URL.
+ reader.readAsText(f);
+ }
}
// load html from file
function LoadFromHtml() {
- var htmlPath = "saves/" + document.getElementById("html-load").value;
- $("#content").load(htmlPath);
+ var htmlPath = "saves/" + document.getElementById("html-load").value;
+ $("#content").load(htmlPath);
- console.log(htmlPath);
+ console.log(htmlPath);
}
document.getElementById('user-project-file').addEventListener('change', handleProjectFileSelect, false);
@@ -258,9 +258,9 @@ document.getElementById('user-project-file').addEventListener('change', handlePr
// activates interactive script after two seconds on page load.
// TODO Load script asynchronously without a the need for a dumb timer
-setTimeout(function(){
- hyper(true);
- // console.log('interactive script');
+setTimeout(function () {
+ hyper(true);
+ // console.log('interactive script');
}, 2000);
// for the audio control (possibly to be deleted eventually)
@@ -273,864 +273,870 @@ var word_start_time;
// load json from url
function getJSONFromUrl() {
- var jsonUrl = document.getElementById("jsonUrll").value;
- console.log("loading json: " + jsonUrl);
- document.getElementById("json-name").innerHTML = jsonUrl;
- // console.log(e);
- // JsonObj = JSON.parse(jsonUrl);
- // console.log(JsonObj);
- // displayTranscript(jsonUrl)
- // $.getJSON(jsonUrl, function (data) {
- // displayTranscript(data);
- // console.log("???");
- // });
- // console.log("poop: " + userJson);
- $.getJSON(jsonUrl, function (data) {
- // displayTranscript(data);
- displayTranscript(data)
- });
+ var jsonUrl = document.getElementById("jsonUrll").value;
+ console.log("loading json: " + jsonUrl);
+ document.getElementById("json-name").innerHTML = jsonUrl;
+ // console.log(e);
+ // JsonObj = JSON.parse(jsonUrl);
+ // console.log(JsonObj);
+ // displayTranscript(jsonUrl)
+ // $.getJSON(jsonUrl, function (data) {
+ // displayTranscript(data);
+ // console.log("???");
+ // });
+ // console.log("poop: " + userJson);
+ $.getJSON(jsonUrl, function (data) {
+ // displayTranscript(data);
+ displayTranscript(data)
+ });
}
-
// display transcript from json file
function displayTranscript(userJson) {
- // get json transcript from user input (default transcript.json)
- // var json = "json/" + document.getElementById("user-filename").value;
- // console.log("loading: " + json);
-
+ // get json transcript from user input (default transcript.json)
+ // var json = "json/" + document.getElementById("user-filename").value;
+ // console.log("loading: " + json);
+
+
+ // $.getJSON(json, function (data) {
+
+ data = userJson
+
+ // assign variables for use in for loop below
+
+ // contains the html to append for each word
+ var text = "";
+ // counts upwards each time the speaker changes, it should be as long as the length of speaker_times
+ var speaker_counter = 0;
+ var new_speaker = "";
+ // counts of how long each para is
+ var paragraphWordCounter = 0;
+ // counts the number of paragraphs
+ var paragraphCounter = 0;
+ var newPara = "";
+
+ // set the number of words after which a new sentence is started
+ // TODO allow user to set
+ var max_para_length = 10;
+
+ let whisper = new WhisperParser( )
+
+ // use the json structure to detect the format being used
+ // eg AWS vs DeepSpeech
+
+ if( whisper.detect( data ) ) {
+
+ whisper.config()
+ whisper.generate( data )
+
+ // parse the DeepSPeech formatted json
+ } else if (data.words) {
+ console.log('Mozilla formatted data detected');
+
+ // turn off confidence toggle
+ document.getElementById('confidence').setAttribute('disabled', 'disabled');
+
+
+ // if (document.getElementById('deepspeech').checked) {
+ var results = data.words
+
+ jsonLength = results.length
+
+
+ transcriptObject = results.words;
+ var confidence = 1;
+
+ // loop through json to appeand words and data
+ // TODO need to adjust this to create a para first
+ // then to append words to that paragraph
+ // then when speaker changes to create a new para
+ for (var i = 0; i < jsonLength; i++) {
+ // get data from JSON string
+
+ var wordLabel = results[i].word;
+ var startTimeLabel = results[i]["start_time "];
+ var durationLabel = results[i].duration;
+
- // $.getJSON(json, function (data) {
+ word = results[i].word;
+ // word start time is in seconds
- data = userJson
+ // create an manual adjustment to data when there is a playback sync error
+ // word highlighting too early means data time is too low, so make it higher
+ // word highlighting too late means data time is too high, so make it lower
+ var adjustment = 0;
+ // var delay = $("#user-delay").val();
+ // // adjustment = Number(delay);
+ // console.log("adjustment: " + adjustment);
- // assign variables for use in for loop below
+ word_start_time = startTimeLabel + adjustment;
+ word_start_time_ms = word_start_time * 1000
- // contains the html to append for each word
- var text = "";
- // counts upwards each time the speaker changes, it should be as long as the length of speaker_times
- var speaker_counter = 0;
- var new_speaker = "";
- // counts of how long each para is
- var paragraphWordCounter = 0;
- // counts the number of paragraphs
- var paragraphCounter = 0;
- var newPara = "";
+ if (results[i + 1] && results[i + 1].start_time) {
+ next_word_start_time = results[i + 1].start_time;
+ // TODO truncaste this as it can go to lots of decimal places
+ // duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
- // set the number of words after which a new sentence is started
- // TODO allow user to set
- var max_para_length = 10;
+ } else if (results[i + 2] && results[i + 2].start_time) {
+ next_word_start_time = results[i + 2].start_time;
+ // TODO truncaste this as it can go to lots of decimal places
+ // duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
+ }
- // use the json structure to detect the format being used
- // eg AWS vs DeepSpeech
+ duration_ms = 1000 * durationLabel;
- // parse the DeepSPeech formatted json
- if (data.words) {
- console.log('Mozilla formatted data detected');
+ if (i == 0) {
- // turn off confidence toggle
- document.getElementById('confidence').setAttribute('disabled', 'disabled');
+ new_speaker = "New Para";
+ // add new para
+ // function takes: timeOfFirstWord, speaker, wordCount
+ paraId = "para-" + paragraphCounter;
+ newPara = CreateNewPara(word_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // document.getElementById('speaker').insertAdjacentHTML('beforebegin',
+ // newPara);
+ };
- // if (document.getElementById('deepspeech').checked) {
- var results = data.words
- jsonLength = results.length
+ // add data to each word: confidence, start time, duration, speaker
+ spanStartTime = "";
+ // create html to be added
+ space = " ";
- transcriptObject = results.words;
- var confidence = 1;
+ text = space + spanStartTime + word + "";
- // loop through json to appeand words and data
- // TODO need to adjust this to create a para first
- // then to append words to that paragraph
- // then when speaker changes to create a new para
- for (var i = 0; i < jsonLength; i++) {
- // get data from JSON string
+ // Uncomment out below to use tooltips
+ // spanTooltip = "";
+ // divTooltip = "";
+ // text = space + divTooltip + spanStartTime + word + "" + spanTooltip + confidence + "
" + word_start_time + "" + "
";
- var wordLabel = results[i].word;
- var startTimeLabel = results[i]["start_time "];
- var durationLabel = results[i].duration;
+ // append text to paragraph
+ para = "#para-" + paragraphCounter;
+ $(para).append(text);
- word = results[i].word;
- // word start time is in seconds
+ // if it gets to a full stop and the current paragraph is too long, start a new paragraph
+ // TODO let user set the paragraph amount
- // create an manual adjustment to data when there is a playback sync error
- // word highlighting too early means data time is too low, so make it higher
- // word highlighting too late means data time is too high, so make it lower
- var adjustment = 0;
- // var delay = $("#user-delay").val();
- // // adjustment = Number(delay);
- // console.log("adjustment: " + adjustment);
- word_start_time = startTimeLabel + adjustment;
- word_start_time_ms = word_start_time * 1000
+ //for (var i = 0; i < speaker_times.length; i++) {
+ //console.log(speaker_times[i]);
+ //}
+ paragraphWordCounter++
- if (results[i + 1] && results[i + 1].start_time) {
- next_word_start_time = results[i + 1].start_time;
- // TODO truncaste this as it can go to lots of decimal places
- // duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
+ if (paragraphWordCounter > max_para_length) {
+ // set data for new speaker
+ paragraphCounter++;
+ paraId = "para-" + paragraphCounter;
+ newPara = CreateNewPara(word_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // reset the paragraph word counter
+ paragraphWordCounter = 0;
+ // console.log(word);
+ // console.log('para too long');
- } else if (results[i + 2] && results[i + 2].start_time) {
- next_word_start_time = results[i + 2].start_time;
- // TODO truncaste this as it can go to lots of decimal places
- // duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
- }
+ };
- duration_ms = 1000 * durationLabel;
+ };
- if (i == 0) {
- new_speaker = "New Para";
- // add new para
- // function takes: timeOfFirstWord, speaker, wordCount
- paraId = "para-" + paragraphCounter;
- newPara = CreateNewPara(word_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // document.getElementById('speaker').insertAdjacentHTML('beforebegin',
- // newPara);
- };
+ } else if (data.results) {
+ // AWS formatted json
+ console.log('AWS formatted data detected');
+ // turn on confidence display toggle
+ document.getElementById('confidence').removeAttribute('disabled');
- // add data to each word: confidence, start time, duration, speaker
- spanStartTime = "";
- // create html to be added
- space = " ";
+ // parse the AWS formatted json
- text = space + spanStartTime + word + "";
+ //
+ var results = data.results;
+ var transcript_raw = JSON.stringify(results.transcripts[0].transcript);
- // Uncomment out below to use tooltips
- // spanTooltip = "";
- // divTooltip = "";
- // text = space + divTooltip + spanStartTime + word + "" + spanTooltip + confidence + "
" + word_start_time + "" + "
";
-
- // append text to paragraph
- para = "#para-" + paragraphCounter;
+ // create empty array to hold speaker names and start times
+ // TODO is this array actually used in anything?
+ // REPLY: yes, it is used to look up who the speaker is depending on the time
+ // Note: in the json a speaker can speak multiple times in a row
+ // we simplify this
- $(para).append(text);
+ if (results.speaker_labels) {
+ console.log('multiple speakers');
+ var whoIsSpeaker
+ var speaker_times = [];
+ var segments = results.speaker_labels.segments;
+ for (var i = 1; i < segments.length; i++) {
+ // check if the speaker has changed
+ if (whoIsSpeaker != segments[i].speaker_label) {
+ // if so add to the array
+ whoIsSpeaker = segments[i].speaker_label
+ speaker = [];
+ speaker.push(segments[i].speaker_label);
+ speaker.push(Number(segments[i].start_time));
+ speaker_times.push(speaker);
+ }
- // if it gets to a full stop and the current paragraph is too long, start a new paragraph
- // TODO let user set the paragraph amount
+ };
+ } else {
+ console.log('one speaker');
+ new_speaker = "speaker";
+ var speaker_times = [[]];
+ var speaker_counter = 0
+ speaker_times[speaker_counter][0] = new_speaker
+ };
+
+
+ // saving global variables for use in audio-control.js (poss can delete)
+ speakerTimes = speaker_times
+
+ transcriptObject = results.items;
+
+ jsonLength = results.items.length;
+
+
+
+ // loop through json to appeand words and data
+ // TODO need to adjust this to create a para first
+ // then to append words to that paragraph
+ // then when speaker changes to create a new para
+ for (var i = 0; i < jsonLength; i++) {
+ // get data from JSON string
+ word = results.items[i].alternatives[0].content;
+ confidence = results.items[i].alternatives[0].confidence;
+ word_start_time = results.items[i].start_time;
+ word_start_time_ms = Math.round(word_start_time * 1000);
+ if (results.items[i + 1] && results.items[i + 1].start_time) {
+ next_word_start_time = results.items[i + 1].start_time;
+ // TODO truncaste this as it can go to lots of decimal places
+ duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
+
+ } else if (results.items[i + 2] && results.items[i + 2].start_time) {
+ next_word_start_time = results.items[i + 2].start_time;
+ ``
+ // TODO truncaste this as it can go to lots of decimal places
+ duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
+ }
+ type = results.items[i].type;
+
+ // check for punctuation and ensure punctuation doesn't have spaces before them
+ if (type == "pronunciation") {
+ space = " ";
+ paragraphWordCounter++;
+ } else if (type == "punctuation") {
+ space = "";
+ };
+
+
+
+ // make sure first word has a speaker - may be unecessary
+ if (i == 0) {
+ // find out and set the speaker counter for the first word
+
+ // // to check who the speaker is at the time of the first word
+ // while (Number(speaker_times[speaker_counter][1]) < Number(word_start_time)) {
+ // speaker_counter++;
+ // };
+
+ // TODO temporary if condition
+ if (results.speaker_labels) {
+ new_speaker = speaker_times[speaker_counter][0];
+ }
- //for (var i = 0; i < speaker_times.length; i++) {
- //console.log(speaker_times[i]);
- //}
- paragraphWordCounter++
+ // add new para
+ // function takes: timeOfFirstWord, speaker, wordCount
+ paraId = "para-" + paragraphCounter;
+ newPara = CreateNewPara(word_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // document.getElementById('speaker').insertAdjacentHTML('beforebegin',
+ // newPara);
+ };
- if (paragraphWordCounter > max_para_length) {
- // set data for new speaker
- paragraphCounter++;
- paraId = "para-" + paragraphCounter;
- newPara = CreateNewPara(word_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // reset the paragraph word counter
- paragraphWordCounter = 0;
- // console.log(word);
- // console.log('para too long');
+ // ok this might need overhaul
+ // want to detect if the speaker has changed
+ // if it has, create a new paragraph and increase speaker counter by 1
+ // speaker_times has the times where the speaker changes
+ // speaker counter starts at 0
+ // when the time of the word exceeds speaker_times 0th row, then change it
- };
-
- };
-
-
-
-
- } else if (data.results) {
- // AWS formatted json
- console.log('AWS formatted data detected');
- // turn on confidence display toggle
- document.getElementById('confidence').removeAttribute('disabled');
-
-
- // parse the AWS formatted json
-
- //
- var results = data.results;
- var transcript_raw = JSON.stringify(results.transcripts[0].transcript);
-
- // create empty array to hold speaker names and start times
- // TODO is this array actually used in anything?
- // REPLY: yes, it is used to look up who the speaker is depending on the time
- // Note: in the json a speaker can speak multiple times in a row
- // we simplify this
-
- if (results.speaker_labels) {
- console.log('multiple speakers');
- var whoIsSpeaker
- var speaker_times = [];
- var segments = results.speaker_labels.segments;
- for (var i = 1; i < segments.length; i++) {
- // check if the speaker has changed
- if (whoIsSpeaker != segments[i].speaker_label) {
- // if so add to the array
- whoIsSpeaker = segments[i].speaker_label
- speaker = [];
- speaker.push(segments[i].speaker_label);
- speaker.push(Number(segments[i].start_time));
- speaker_times.push(speaker);
- }
-
- };
- } else {
- console.log('one speaker');
- new_speaker = "speaker";
- var speaker_times = [[]];
- var speaker_counter = 0
- speaker_times[speaker_counter][0] = new_speaker
- };
-
-
- // saving global variables for use in audio-control.js (poss can delete)
- speakerTimes = speaker_times
-
- transcriptObject = results.items;
-
- jsonLength = results.items.length;
-
-
-
- // loop through json to appeand words and data
- // TODO need to adjust this to create a para first
- // then to append words to that paragraph
- // then when speaker changes to create a new para
- for (var i = 0; i < jsonLength; i++) {
- // get data from JSON string
- word = results.items[i].alternatives[0].content;
- confidence = results.items[i].alternatives[0].confidence;
- word_start_time = results.items[i].start_time;
- word_start_time_ms = Math.round(word_start_time * 1000);
- if (results.items[i + 1] && results.items[i + 1].start_time) {
- next_word_start_time = results.items[i + 1].start_time;
- // TODO truncaste this as it can go to lots of decimal places
- duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
-
- } else if (results.items[i + 2] && results.items[i + 2].start_time) {
- next_word_start_time = results.items[i + 2].start_time;
- ``
- // TODO truncaste this as it can go to lots of decimal places
- duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
- }
- type = results.items[i].type;
-
- // check for punctuation and ensure punctuation doesn't have spaces before them
- if (type == "pronunciation") {
- space = " ";
- paragraphWordCounter++;
- } else if (type == "punctuation") {
- space = "";
- };
-
-
-
- // make sure first word has a speaker - may be unecessary
- if (i == 0) {
- // find out and set the speaker counter for the first word
-
- // // to check who the speaker is at the time of the first word
- // while (Number(speaker_times[speaker_counter][1]) < Number(word_start_time)) {
- // speaker_counter++;
- // };
-
- // TODO temporary if condition
- if (results.speaker_labels) {
- new_speaker = speaker_times[speaker_counter][0];
- }
-
-
- // add new para
- // function takes: timeOfFirstWord, speaker, wordCount
- paraId = "para-" + paragraphCounter;
- newPara = CreateNewPara(word_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // document.getElementById('speaker').insertAdjacentHTML('beforebegin',
- // newPara);
- };
-
- // ok this might need overhaul
- // want to detect if the speaker has changed
- // if it has, create a new paragraph and increase speaker counter by 1
- // speaker_times has the times where the speaker changes
- // speaker counter starts at 0
- // when the time of the word exceeds speaker_times 0th row, then change it
-
- // add new para if speaker changes
- // checks if it's not the last speaker
-
- // TODO temporary if condition
- if (results.speaker_labels) {
- if ((speaker_counter < speaker_times.length) && (i != 0)) {
-
- speakerStart = speaker_times[speaker_counter][1]
- // checks if the time of the speaker is less than the current word
- // ok to do this, we need to check for the next word, not this one
- // also what if the next word is punctuation
- if (speakerStart < next_word_start_time) {
-
-
- // checks if the amount of time the speaker spoke for is more than a second
- // might be able to remove this since it addressed a problems that's been solved elsewhere
- var min_time = 1;
- // if
- if (speaker_times[speaker_counter + 1] && (speaker_times[speaker_counter + 1][1] - speaker_times[speaker_counter][1] > min_time)) {
- speaker_counter++;
- // checks if the speaker has changed
- if (new_speaker != speaker_times[speaker_counter][0]) {
- // console.log(speaker_times);
-
- // console.log(word);
- // console.log(speaker_counter);
-
-
- // changes the speaker variable
- new_speaker = speaker_times[speaker_counter][0];
-
- // add a new para
- paragraphCounter++;
- paraId = "para-" + paragraphCounter;
-
- newPara = CreateNewPara(word_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // reset the paragraph word counter
- paragraphWordCounter = 0;
-
-
-
- };
- };
-
- };
- };
- }
-
-
-
- // add data to each word: confidence, start time, duration, speaker
- spanStartTime = "";
- // create html to be added
-
-
-
- if (type == "pronunciation") {
- // remove
-
- text = space + spanStartTime + word + "";
- } else if (type == "punctuation") {
- // check if the previous word was also punctuation cause by removing an utterence
-
- text = space + word
-
-
- };
-
- // Uncomment out below to use tooltips
- // spanTooltip = "";
- // divTooltip = "";
- // text = space + divTooltip + spanStartTime + word + "" + spanTooltip + confidence + "
" + word_start_time + "" + "
";
-
- // append text to paragraph
- para = "#para-" + paragraphCounter;
-
- $(para).append(text);
-
- // if it gets to a full stop and the current paragraph is too long, start a new paragraph
- // TODO let user set the paragraph amount
-
-
-
-
- if (type == "punctuation" && (word == "." || word == "!" || word == "?") && paragraphWordCounter > max_para_length && new_speaker == speaker_times[speaker_counter][0]) {
- // set data for new speaker
- paragraphCounter++;
- paraId = "para-" + paragraphCounter;
-
- // use next word start time as current one is punctuation
- newPara = CreateNewPara(next_word_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // reset the paragraph word counter
- paragraphWordCounter = 0;
- // console.log(word);
- // console.log('para too long');
-
- };
-
- //for (var i = 0; i < speaker_times.length; i++) {
- //console.log(speaker_times[i]);
- //}
-
- };
-
-
- } else if (data.monologues) {
- // rev.ai formatted json
- console.log('rev.ai formatted data detected');
- // turn on confidence display toggle
- document.getElementById('confidence').removeAttribute('disabled');
-
- // parse the rev.ai formatted json
-
- // rev.ai json structure has everything in "monologues"
- // it includes spaces as punctuation. It has confidence values for each word.
- // speakers are declared in new objects in monologues
- // start and end time are in seconds
-
- // console.log("length: " + data.monologues.length);
-
- var spearkerchanges = data.monologues.length;
+ // add new para if speaker changes
+ // checks if it's not the last speaker
- // loop through all the speaker changes
- for (var j = 0; j < spearkerchanges; j++) {
- var results = data.monologues[j];
+ // TODO temporary if condition
+ if (results.speaker_labels) {
+ if ((speaker_counter < speaker_times.length) && (i != 0)) {
- // create empty array to hold speaker names and start times
- // var whoIsSpeaker
- // var speaker_times = [];
- // var segments = results.speaker_labels.segments;
- // for (var i = 1; i < segments.length; i++) {
- // // check if the speaker has changed
- // if (whoIsSpeaker != segments[i].speaker_label) {
- // // if so add to the array
- // whoIsSpeaker = segments[i].speaker_label
- // speaker = [];
- // speaker.push(segments[i].speaker_label);
- // speaker.push(Number(segments[i].start_time));
- // speaker_times.push(speaker);
- // }
+ speakerStart = speaker_times[speaker_counter][1]
+ // checks if the time of the speaker is less than the current word
+ // ok to do this, we need to check for the next word, not this one
+ // also what if the next word is punctuation
+ if (speakerStart < next_word_start_time) {
- // };
- // saving global variables for use in audio-control.js
- // TODO find another way without global variables
- speakerTimes = speaker_times
- transcriptObject = results.items;
- jsonLength = results.elements.length;
+ // checks if the amount of time the speaker spoke for is more than a second
+ // might be able to remove this since it addressed a problems that's been solved elsewhere
+ var min_time = 1;
+ // if
+ if (speaker_times[speaker_counter + 1] && (speaker_times[speaker_counter + 1][1] - speaker_times[speaker_counter][1] > min_time)) {
+ speaker_counter++;
+ // checks if the speaker has changed
+ if (new_speaker != speaker_times[speaker_counter][0]) {
+ // console.log(speaker_times);
- // first, let's just try one speaker.
- // Next deal with more
- // Maybe loop through: get length of each elements array
+ // console.log(word);
+ // console.log(speaker_counter);
- // loop through json to appeand words and data
+ // changes the speaker variable
+ new_speaker = speaker_times[speaker_counter][0];
- para_start_time = results.elements[0].ts;
- new_speaker = results.speaker
- console.log("speaker: " + new_speaker + " start time: " + para_start_time + " paraId: " + paragraphCounter);
+ // add a new para
paragraphCounter++;
+ paraId = "para-" + paragraphCounter;
+
+ newPara = CreateNewPara(word_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // reset the paragraph word counter
+ paragraphWordCounter = 0;
+
+
+
+ };
+ };
+
+ };
+ };
+ }
+
+
+
+ // add data to each word: confidence, start time, duration, speaker
+ spanStartTime = "";
+ // create html to be added
+
+
+
+ if (type == "pronunciation") {
+ // remove
- // add first paragraph when speaker changes
-
-
-
- // find out and set the speaker counter for the first word
-
- // // to check who the speaker is at the time of the first word
- // while (Number(speaker_times[speaker_counter][1]) < Number(word_start_time)) {
- // speaker_counter++;
- // };
- // new_speaker = speaker_times[speaker_counter][0];
-
- // add new para
- // function takes: timeOfFirstWord, speaker, wordCount
- paraId = "para-" + paragraphCounter;
- newPara = CreateNewPara(para_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // document.getElementById('speaker').insertAdjacentHTML('beforebegin',
- // newPara);
-
-
- // loop through all words for each speaker
- for (var i = 0; i < jsonLength; i++) {
- // define word data from JSON string
- word = results.elements[i].value;
- confidence = results.elements[i].confidence;
- word_start_time = results.elements[i].ts;
- word_start_time_ms = Math.round(word_start_time * 1000);
- word_end_time = results.elements[i].end_ts;
- duration_ms = duration_ms = Math.round(1000 * (word_end_time - word_start_time))
- type = results.elements[i].type;
-
- // work out next word start time and duration (for new paragraph tag)
- if (results.elements[i + 1] && results.elements[i + 1].ts) {
- next_word_start_time = results.elements[i + 1].ts;
- // TODO truncaste this as it can go to lots of decimal places
- duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
-
- } else if (results.elements[i + 2] && results.elements[i + 2].ts) {
- next_word_start_time = results.elements[i + 2].ts;
- ``
- // TODO truncaste this as it can go to lots of decimal places
- duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
- } else if (results.elements[i + 3] && results.elements[i + 3].ts) {
- next_word_start_time = results.elements[i + 2].ts;
- ``
- // TODO truncaste this as it can go to lots of decimal places
- duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
- }
-
-
- // check for punctuation and ensure punctuation doesn't have spaces before them
- // if (type == "pronunciation") {
- // space = " ";
- // paragraphWordCounter++;
- // } else if (type == "punctuation") {
- // space = "";
- // };
-
- // ok this might need overhaul
- // want to detect if the speaker has changed
- // if it has, create a new paragraph and increase speaker counter by 1
- // speaker_times has the times where the speaker changes
- // speaker counter starts at 0
- // when the time of the word exceeds speaker_times 0th row, then change it
-
- // add new para if speaker changes
- // checks if it's not the last speaker
- // this only works for AWS
-
- // if ((speaker_counter < speaker_times.length) && (i != 0)) {
-
- // speakerStart = speaker_times[speaker_counter][1]
- // // checks if the time of the speaker is less than the current word
- // // ok to do this, we need to check for the next word, not this one
- // // also what if the next word is punctuation
- // if (speakerStart < next_word_start_time) {
-
-
- // // checks if the amount of time the speaker spoke for is more than a second
- // // might be able to remove this since it addressed a problems that's been solved elsewhere
- // var min_time = 1;
- // // if
- // if (speaker_times[speaker_counter + 1] && (speaker_times[speaker_counter + 1][1] - speaker_times[speaker_counter][1] > min_time)) {
- // speaker_counter++;
- // // checks if the speaker has changed
- // if (new_speaker != speaker_times[speaker_counter][0]) {
- // // console.log(speaker_times);
-
- // // console.log(word);
- // // console.log(speaker_counter);
-
-
- // // changes the speaker variable
- // new_speaker = speaker_times[speaker_counter][0];
-
- // // add a new para
- // paragraphCounter++;
- // paraId = "para-" + paragraphCounter;
-
- // newPara = CreateNewPara(word_start_time, new_speaker, paraId);
- // $('#content').append(newPara);
- // // reset the paragraph word counter
- // paragraphWordCounter = 0;
-
-
-
- // };
- // };
-
- // };
- // };
-
- // remove uh and um
- if (type =="text" && (word == "uh" || word == "um")) {
- if (results.elements[i + 1].type == "punct") {
- i++;
- }
- } else {
- // add data to each word: confidence, start time, duration, speaker
- spanStartTime = "";
- // create html to be added
-
- // add words and punctuation
- if (type == "text") {
- text = spanStartTime + word + "";
- paragraphWordCounter++;
- } else if (type == "punct") {
- // check if the previous word was also punctuation cause by removing an utterence
- text = word
- };
-
- // Uncomment out below to use tooltips
- // spanTooltip = "";
- // divTooltip = "";
- // text = space + divTooltip + spanStartTime + word + "" + spanTooltip + confidence + "
" + word_start_time + "" + "
";
-
- // append text to paragraph
- para = "#para-" + paragraphCounter;
-
- $(para).append(text);
+ text = space + spanStartTime + word + "";
+ } else if (type == "punctuation") {
+ // check if the previous word was also punctuation cause by removing an utterence
- // if it gets to a full stop and the current paragraph is too long, start a new paragraph
- // TODO let user set the paragraph amount
+ text = space + word
- // console.log(i + " : " + paragraphWordCounter);
- // start a new paragraph if text current length is over a max threshold and it reaches the end of a sentence
- if (type == "punct" && (word == "." || word == "!" || word == "?") && paragraphWordCounter > max_para_length && i < jsonLength - 2) {
- // set data for new speaker
-
- paragraphCounter++;
- paraId = "para-" + paragraphCounter;
+ };
- // use next word start time as current one is punctuation
- // new_speaker = data.monologues[j + 1].speaker
- newPara = CreateNewPara(next_word_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // reset the paragraph word counter
- paragraphWordCounter = 0;
- // console.log(word);
- // console.log('para too long');
- };
+ // Uncomment out below to use tooltips
+ // spanTooltip = "";
+ // divTooltip = "";
+ // text = space + divTooltip + spanStartTime + word + "" + spanTooltip + confidence + "
" + word_start_time + "" + "
";
- //for (var i = 0; i < speaker_times.length; i++) {
- //console.log(speaker_times[i]);
- //}
- }
- };
- }
+ // append text to paragraph
+ para = "#para-" + paragraphCounter;
- } else if (data.response.chunks) {
- // Yandex formatted json
- // TODO test this
+ $(para).append(text);
- console.log('YandexSpeechKit formatted data detected');
- // turn on confidence display toggle
- document.getElementById('confidence').removeAttribute('disabled');
+ // if it gets to a full stop and the current paragraph is too long, start a new paragraph
+ // TODO let user set the paragraph amount
- // turn off confidence toggle
- document.getElementById('confidence').setAttribute('disabled', 'disabled');
- var results = data.response.chunks
- jsonLength = results.length
+ if (type == "punctuation" && (word == "." || word == "!" || word == "?") && paragraphWordCounter > max_para_length && new_speaker == speaker_times[speaker_counter][0]) {
+ // set data for new speaker
+ paragraphCounter++;
+ paraId = "para-" + paragraphCounter;
+ // use next word start time as current one is punctuation
+ newPara = CreateNewPara(next_word_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // reset the paragraph word counter
+ paragraphWordCounter = 0;
+ // console.log(word);
+ // console.log('para too long');
- transcriptObject = results.words;
- var confidence = 1;
+ };
- // loop through json to appeand words and data
- // TODO need to adjust this to create a para first
- // then to append words to that paragraph
- // then when speaker changes to create a new para
- for (var j = 0; j < jsonLength; j++) {
- if (results[j].channelTag == '1') {
- chunk = results[j].alternatives[0].words;
- chunkLength = chunk.length;
+ //for (var i = 0; i < speaker_times.length; i++) {
+ //console.log(speaker_times[i]);
+ //}
- if (j == 0) {
+ };
- new_speaker = "New Para";
- // add new para
- // function takes: timeOfFirstWord, speaker, wordCount
- paraId = "para-" + paragraphCounter;
- newPara = CreateNewPara('0', new_speaker, paraId);
- $('#content').append(newPara);
+ } else if (data.monologues) {
+ // rev.ai formatted json
+ console.log('rev.ai formatted data detected');
+ // turn on confidence display toggle
+ document.getElementById('confidence').removeAttribute('disabled');
- };
+ // parse the rev.ai formatted json
- for (var i = 0; i < chunkLength; i++) {
- // get data from JSON string
- // As yandex split text by pauses let assume that each fragment is a sentence.
- if (i==0){
- chunk[i].word = chunk[i].word[0].toUpperCase() + chunk[i].word.substring(1);
- }else if (i == chunkLength-1){
- chunk[i].word = chunk[i].word + '.';
- }
- wordLabel = chunk[i].word;
- startTimeLabel = Number(chunk[i]["startTime"].substr(0, chunk[i]["startTime"].length - 1));
- durationLabel = Number(chunk[i]["endTime"].substr(0, chunk[i]["endTime"].length - 1)) - Number(chunk[i]["startTime"].substr(0, chunk[i]["startTime"].length - 1));
+ // rev.ai json structure has everything in "monologues"
+ // it includes spaces as punctuation. It has confidence values for each word.
+ // speakers are declared in new objects in monologues
+ // start and end time are in seconds
+ // console.log("length: " + data.monologues.length);
- word = chunk[i].word;
- // word start time is in seconds
+ var spearkerchanges = data.monologues.length;
- // create an manual adjustment to data when there is a playback sync error
- // word highlighting too early means data time is too low, so make it higher
- // word highlighting too late means data time is too high, so make it lower
- var adjustment = 0;
+ // loop through all the speaker changes
+ for (var j = 0; j < spearkerchanges; j++) {
+ var results = data.monologues[j];
- word_start_time = startTimeLabel + adjustment;
- word_start_time_ms = word_start_time * 1000
+ // create empty array to hold speaker names and start times
+ // var whoIsSpeaker
+ // var speaker_times = [];
+ // var segments = results.speaker_labels.segments;
+ // for (var i = 1; i < segments.length; i++) {
+ // // check if the speaker has changed
+ // if (whoIsSpeaker != segments[i].speaker_label) {
+ // // if so add to the array
+ // whoIsSpeaker = segments[i].speaker_label
+ // speaker = [];
+ // speaker.push(segments[i].speaker_label);
+ // speaker.push(Number(segments[i].start_time));
+ // speaker_times.push(speaker);
+ // }
+ // };
- duration_ms = 1000 * durationLabel;
+ // saving global variables for use in audio-control.js
+ // TODO find another way without global variables
+ speakerTimes = speaker_times
+ transcriptObject = results.items;
+ jsonLength = results.elements.length;
+ // first, let's just try one speaker.
+ // Next deal with more
+ // Maybe loop through: get length of each elements array
+ // loop through json to appeand words and data
- // add data to each word: confidence, start time, duration, speaker
- spanStartTime = "";
- // create html to be added
+ para_start_time = results.elements[0].ts;
+ new_speaker = results.speaker
+ console.log("speaker: " + new_speaker + " start time: " + para_start_time + " paraId: " + paragraphCounter);
+ paragraphCounter++;
- space = " ";
+ // add first paragraph when speaker changes
- text = space + spanStartTime + word + "";
- // append text to paragraph
- para = "#para-" + paragraphCounter;
- $(para).append(text);
+ // find out and set the speaker counter for the first word
- paragraphWordCounter++
+ // // to check who the speaker is at the time of the first word
+ // while (Number(speaker_times[speaker_counter][1]) < Number(word_start_time)) {
+ // speaker_counter++;
+ // };
+ // new_speaker = speaker_times[speaker_counter][0];
+
+ // add new para
+ // function takes: timeOfFirstWord, speaker, wordCount
+ paraId = "para-" + paragraphCounter;
+ newPara = CreateNewPara(para_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // document.getElementById('speaker').insertAdjacentHTML('beforebegin',
+ // newPara);
+
+
+ // loop through all words for each speaker
+ for (var i = 0; i < jsonLength; i++) {
+ // define word data from JSON string
+ word = results.elements[i].value;
+ confidence = results.elements[i].confidence;
+ word_start_time = results.elements[i].ts;
+ word_start_time_ms = Math.round(word_start_time * 1000);
+ word_end_time = results.elements[i].end_ts;
+ duration_ms = duration_ms = Math.round(1000 * (word_end_time - word_start_time))
+ type = results.elements[i].type;
+
+ // work out next word start time and duration (for new paragraph tag)
+ if (results.elements[i + 1] && results.elements[i + 1].ts) {
+ next_word_start_time = results.elements[i + 1].ts;
+ // TODO truncaste this as it can go to lots of decimal places
+ duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
+
+ } else if (results.elements[i + 2] && results.elements[i + 2].ts) {
+ next_word_start_time = results.elements[i + 2].ts;
+ ``
+ // TODO truncaste this as it can go to lots of decimal places
+ duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
+ } else if (results.elements[i + 3] && results.elements[i + 3].ts) {
+ next_word_start_time = results.elements[i + 2].ts;
+ ``
+ // TODO truncaste this as it can go to lots of decimal places
+ duration_ms = Math.round(1000 * (next_word_start_time - word_start_time))
+ }
- };
- // new para each 4 chunks
- if ((j % 4) == 0) {
- paragraphCounter++;
- paraId = "para-" + paragraphCounter;
- newPara = CreateNewPara(word_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // reset the paragraph word counter
- paragraphWordCounter = 0;
- // console.log(word);
- // console.log('para too long');
- }
- }
- }
- }
+ // check for punctuation and ensure punctuation doesn't have spaces before them
+ // if (type == "pronunciation") {
+ // space = " ";
+ // paragraphWordCounter++;
+ // } else if (type == "punctuation") {
+ // space = "";
+ // };
+
+ // ok this might need overhaul
+ // want to detect if the speaker has changed
+ // if it has, create a new paragraph and increase speaker counter by 1
+ // speaker_times has the times where the speaker changes
+ // speaker counter starts at 0
+ // when the time of the word exceeds speaker_times 0th row, then change it
+
+ // add new para if speaker changes
+ // checks if it's not the last speaker
+ // this only works for AWS
+
+ // if ((speaker_counter < speaker_times.length) && (i != 0)) {
+
+ // speakerStart = speaker_times[speaker_counter][1]
+ // // checks if the time of the speaker is less than the current word
+ // // ok to do this, we need to check for the next word, not this one
+ // // also what if the next word is punctuation
+ // if (speakerStart < next_word_start_time) {
+
+
+ // // checks if the amount of time the speaker spoke for is more than a second
+ // // might be able to remove this since it addressed a problems that's been solved elsewhere
+ // var min_time = 1;
+ // // if
+ // if (speaker_times[speaker_counter + 1] && (speaker_times[speaker_counter + 1][1] - speaker_times[speaker_counter][1] > min_time)) {
+ // speaker_counter++;
+ // // checks if the speaker has changed
+ // if (new_speaker != speaker_times[speaker_counter][0]) {
+ // // console.log(speaker_times);
+
+ // // console.log(word);
+ // // console.log(speaker_counter);
+
+
+ // // changes the speaker variable
+ // new_speaker = speaker_times[speaker_counter][0];
+
+ // // add a new para
+ // paragraphCounter++;
+ // paraId = "para-" + paragraphCounter;
+
+ // newPara = CreateNewPara(word_start_time, new_speaker, paraId);
+ // $('#content').append(newPara);
+ // // reset the paragraph word counter
+ // paragraphWordCounter = 0;
+
+
+
+ // };
+ // };
+
+ // };
+ // };
+
+ // remove uh and um
+ if (type == "text" && (word == "uh" || word == "um")) {
+ if (results.elements[i + 1].type == "punct") {
+ i++;
+ }
+ } else {
+ // add data to each word: confidence, start time, duration, speaker
+ spanStartTime = "";
+ // create html to be added
+
+ // add words and punctuation
+ if (type == "text") {
+ text = spanStartTime + word + "";
+ paragraphWordCounter++;
+ } else if (type == "punct") {
+ // check if the previous word was also punctuation cause by removing an utterence
+ text = word
+ };
+
+ // Uncomment out below to use tooltips
+ // spanTooltip = "";
+ // divTooltip = "";
+ // text = space + divTooltip + spanStartTime + word + "" + spanTooltip + confidence + "
" + word_start_time + "" + "
";
+
+ // append text to paragraph
+ para = "#para-" + paragraphCounter;
+
+ $(para).append(text);
+
+ // if it gets to a full stop and the current paragraph is too long, start a new paragraph
+ // TODO let user set the paragraph amount
+
+ // console.log(i + " : " + paragraphWordCounter);
+
+ // start a new paragraph if text current length is over a max threshold and it reaches the end of a sentence
+ if (type == "punct" && (word == "." || word == "!" || word == "?") && paragraphWordCounter > max_para_length && i < jsonLength - 2) {
+ // set data for new speaker
+
+ paragraphCounter++;
+ paraId = "para-" + paragraphCounter;
+
+ // use next word start time as current one is punctuation
+ // new_speaker = data.monologues[j + 1].speaker
+ newPara = CreateNewPara(next_word_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // reset the paragraph word counter
+ paragraphWordCounter = 0;
+ // console.log(word);
+ // console.log('para too long');
+ };
+
+ //for (var i = 0; i < speaker_times.length; i++) {
+ //console.log(speaker_times[i]);
+ //}
+ }
+ };
+ }
- // Google SpeechKit json parser (Works only if speaker diarization turned on!)
- // TODO test this
- else if (data.response.results) {
- console.log('GoogleSpeechKit formatted data detected');
- // turn on confidence display toggle
- document.getElementById('confidence').removeAttribute('disabled');
+ } else if (data.response.chunks) {
+ // Yandex formatted json
+ // TODO test this
- // turn off confidence toggle
- document.getElementById('confidence').setAttribute('disabled', 'disabled');
+ console.log('YandexSpeechKit formatted data detected');
+ // turn on confidence display toggle
+ document.getElementById('confidence').removeAttribute('disabled');
+ // turn off confidence toggle
+ document.getElementById('confidence').setAttribute('disabled', 'disabled');
- var results = data.response.results
- jsonLength = results.length
+ var results = data.response.chunks
- if (results[jsonLength-1].alternatives[0].words[0]['speakerTag'] != undefined){
- results = [results[jsonLength-1]]
- jsonLength = results.length
- }
+ jsonLength = results.length
- var confidence = 1;
- // loop through json to appeand words and data
- // TODO need to adjust this to create a para first
- // then to append words to that paragraph
- // then when speaker changes to create a new para
- for (var j = 0; j < jsonLength; j++) {
- chunk = results[j].alternatives[0].words;
- chunkLength = chunk.length;
+ transcriptObject = results.words;
+ var confidence = 1;
- if (j == 0) {
+ // loop through json to appeand words and data
+ // TODO need to adjust this to create a para first
+ // then to append words to that paragraph
+ // then when speaker changes to create a new para
+ for (var j = 0; j < jsonLength; j++) {
+ if (results[j].channelTag == '1') {
+ chunk = results[j].alternatives[0].words;
+ chunkLength = chunk.length;
- new_speaker = "spk_"+chunk[0].speakerTag;
+ if (j == 0) {
- // add new para
- // function takes: timeOfFirstWord, speaker, wordCount
- paraId = "para-" + paragraphCounter;
- newPara = CreateNewPara('0', new_speaker, paraId);
- $('#content').append(newPara);
+ new_speaker = "New Para";
- };
+ // add new para
+ // function takes: timeOfFirstWord, speaker, wordCount
+ paraId = "para-" + paragraphCounter;
+ newPara = CreateNewPara('0', new_speaker, paraId);
+ $('#content').append(newPara);
- for (var i = 0; i < chunkLength; i++) {
- // get data from JSON string
+ };
- wordLabel = chunk[i].word;
- startTimeLabel = Number(chunk[i]["startTime"].substr(0, chunk[i]["startTime"].length - 1));
- durationLabel = Number(chunk[i]["endTime"].substr(0, chunk[i]["endTime"].length - 1)) - Number(chunk[i]["startTime"].substr(0, chunk[i]["startTime"].length - 1));
+ for (var i = 0; i < chunkLength; i++) {
+ // get data from JSON string
+ // As yandex split text by pauses let assume that each fragment is a sentence.
+ if (i == 0) {
+ chunk[i].word = chunk[i].word[0].toUpperCase() + chunk[i].word.substring(1);
+ } else if (i == chunkLength - 1) {
+ chunk[i].word = chunk[i].word + '.';
+ }
+ wordLabel = chunk[i].word;
+ startTimeLabel = Number(chunk[i]["startTime"].substr(0, chunk[i]["startTime"].length - 1));
+ durationLabel = Number(chunk[i]["endTime"].substr(0, chunk[i]["endTime"].length - 1)) - Number(chunk[i]["startTime"].substr(0, chunk[i]["startTime"].length - 1));
- if (i < chunkLength-1){
- nextStartTime = Number(chunk[i+1]["startTime"].substr(0, chunk[i]["startTime"].length - 1))-Number(chunk[i]["endTime"].substr(0, chunk[i]["endTime"].length - 1));
- }else{
- nextStartTime = 0;
- }
- word = chunk[i].word;
- // word start time is in seconds
+ word = chunk[i].word;
+ // word start time is in seconds
- // create an manual adjustment to data when there is a playback sync error
- // word highlighting too early means data time is too low, so make it higher
- // word highlighting too late means data time is too high, so make it lower
- var adjustment = 0;
+ // create an manual adjustment to data when there is a playback sync error
+ // word highlighting too early means data time is too low, so make it higher
+ // word highlighting too late means data time is too high, so make it lower
+ var adjustment = 0;
- word_start_time = startTimeLabel + adjustment;
- word_start_time_ms = word_start_time * 1000
+ word_start_time = startTimeLabel + adjustment;
+ word_start_time_ms = word_start_time * 1000
- next_word_start_time = nextStartTime + adjustment;
- next_word_start_time_ms = next_word_start_time * 1000
- duration_ms = 1000 * durationLabel;
+ duration_ms = 1000 * durationLabel;
- // add data to each word: confidence, start time, duration, speaker
- spanStartTime = "";
- // create html to be added
+ // add data to each word: confidence, start time, duration, speaker
+ spanStartTime = "";
+ // create html to be added
- space = " ";
+ space = " ";
- text = space + spanStartTime + word + "";
+ text = space + spanStartTime + word + "";
- // append text to paragraph
- para = "#para-" + paragraphCounter;
+ // append text to paragraph
+ para = "#para-" + paragraphCounter;
- $(para).append(text);
+ $(para).append(text);
+
+ paragraphWordCounter++
+
+ };
+ // new para each 4 chunks
+ if ((j % 4) == 0) {
+ paragraphCounter++;
+ paraId = "para-" + paragraphCounter;
+ newPara = CreateNewPara(word_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // reset the paragraph word counter
+ paragraphWordCounter = 0;
+ // console.log(word);
+ // console.log('para too long');
+ }
- paragraphWordCounter++
+ }
+ }
+ }
- //if (paragraphWordCounter > max_para_length) {
- // set data for new speaker
+ // Google SpeechKit json parser (Works only if speaker diarization turned on!)
+ // TODO test this
+ else if (data.response.results) {
+ console.log('GoogleSpeechKit formatted data detected');
+ // turn on confidence display toggle
+ document.getElementById('confidence').removeAttribute('disabled');
- if (i < chunkLength-2 && chunk[i].speakerTag != chunk[i+1].speakerTag && chunk[i].speakerTag != chunk[i+2].speakerTag){
- paragraphCounter++;
- paraId = "para-" + paragraphCounter;
- new_speaker = 'spk_'+chunk[i+1].speakerTag
- newPara = CreateNewPara(word_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // reset the paragraph word counter
- paragraphWordCounter = 0;
- }
- //};
+ // turn off confidence toggle
+ document.getElementById('confidence').setAttribute('disabled', 'disabled');
- };
+ var results = data.response.results
+ jsonLength = results.length
- if ((j % 4) == 0) {
- paragraphCounter++;
- paraId = "para-" + paragraphCounter;
- newPara = CreateNewPara(word_start_time, new_speaker, paraId);
- $('#content').append(newPara);
- // reset the paragraph word counter
- paragraphWordCounter = 0;
- }
+ if (results[jsonLength - 1].alternatives[0].words[0]['speakerTag'] != undefined) {
+ results = [results[jsonLength - 1]]
+ jsonLength = results.length
+ }
+ var confidence = 1;
- }
- }
+ // loop through json to appeand words and data
+ // TODO need to adjust this to create a para first
+ // then to append words to that paragraph
+ // then when speaker changes to create a new para
+ for (var j = 0; j < jsonLength; j++) {
+ chunk = results[j].alternatives[0].words;
+ chunkLength = chunk.length;
+ if (j == 0) {
- var obj = JSON.stringify(results);
- $('.raw').html(transcript_raw);
- $('.whole').html(obj);
+ new_speaker = "spk_" + chunk[0].speakerTag;
- // });
+ // add new para
+ // function takes: timeOfFirstWord, speaker, wordCount
+ paraId = "para-" + paragraphCounter;
+ newPara = CreateNewPara('0', new_speaker, paraId);
+ $('#content').append(newPara);
+ };
- var autoScrollCheck = document.getElementById("autoscroll-off").checked;
- if (autoScrollCheck) {
- setTimeout(
- function () {
- console.log('transcript being initiated');
- hyper(true);
+ for (var i = 0; i < chunkLength; i++) {
+ // get data from JSON string
- }, 1000)
+ wordLabel = chunk[i].word;
+ startTimeLabel = Number(chunk[i]["startTime"].substr(0, chunk[i]["startTime"].length - 1));
+ durationLabel = Number(chunk[i]["endTime"].substr(0, chunk[i]["endTime"].length - 1)) - Number(chunk[i]["startTime"].substr(0, chunk[i]["startTime"].length - 1));
+
+ if (i < chunkLength - 1) {
+ nextStartTime = Number(chunk[i + 1]["startTime"].substr(0, chunk[i]["startTime"].length - 1)) - Number(chunk[i]["endTime"].substr(0, chunk[i]["endTime"].length - 1));
+ } else {
+ nextStartTime = 0;
}
+ word = chunk[i].word;
+ // word start time is in seconds
+
+ // create an manual adjustment to data when there is a playback sync error
+ // word highlighting too early means data time is too low, so make it higher
+ // word highlighting too late means data time is too high, so make it lower
+ var adjustment = 0;
+
+ word_start_time = startTimeLabel + adjustment;
+ word_start_time_ms = word_start_time * 1000
+
+ next_word_start_time = nextStartTime + adjustment;
+ next_word_start_time_ms = next_word_start_time * 1000
+
+ duration_ms = 1000 * durationLabel;
+
+
+
+
+ // add data to each word: confidence, start time, duration, speaker
+ spanStartTime = "";
+ // create html to be added
+
+ space = " ";
+
+ text = space + spanStartTime + word + "";
+
+ // append text to paragraph
+ para = "#para-" + paragraphCounter;
+
+ $(para).append(text);
+
+ paragraphWordCounter++
+
+ //if (paragraphWordCounter > max_para_length) {
+ // set data for new speaker
+
+ if (i < chunkLength - 2 && chunk[i].speakerTag != chunk[i + 1].speakerTag && chunk[i].speakerTag != chunk[i + 2].speakerTag) {
+ paragraphCounter++;
+ paraId = "para-" + paragraphCounter;
+ new_speaker = 'spk_' + chunk[i + 1].speakerTag
+ newPara = CreateNewPara(word_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // reset the paragraph word counter
+ paragraphWordCounter = 0;
+ }
+ //};
+
+
+ };
+
+
+ if ((j % 4) == 0) {
+ paragraphCounter++;
+ paraId = "para-" + paragraphCounter;
+ newPara = CreateNewPara(word_start_time, new_speaker, paraId);
+ $('#content').append(newPara);
+ // reset the paragraph word counter
+ paragraphWordCounter = 0;
+ }
+
+
+ }
+ }
+
+
+ var obj = JSON.stringify(results);
+ $('.raw').html(transcript_raw);
+ $('.whole').html(obj);
+
+ // });
+
+
+ var autoScrollCheck = document.getElementById("autoscroll-off").checked;
+ if (autoScrollCheck) {
+ setTimeout(
+ function () {
+ console.log('transcript being initiated');
+ hyper(true);
+
+ }, 1000)
+ }
+
};
// Link
@@ -1138,19 +1144,19 @@ function displayTranscript(userJson) {
// copy to clipboard
function CopyToClipboard(containerid) {
- // Create a new textarea element and give it id='temp_element'
- var textarea = document.createElement('textarea');
- textarea.id = 'temp_element';
- // Optional step to make less noise on the page, if any!
- textarea.style.height = 0;
- // Now append it to your page somewhere, I chose
- document.body.appendChild(textarea);
- // Give our textarea a value of whatever inside the div of id=containerid
- textarea.value = document.getElementById(containerid).innerText;
- // Now copy whatever inside the textarea to clipboard
- var selector = document.querySelector('#temp_element');
- selector.select();
- document.execCommand('copy');
- // Remove the textarea
- document.body.removeChild(textarea);
+ // Create a new textarea element and give it id='temp_element'
+ var textarea = document.createElement('textarea');
+ textarea.id = 'temp_element';
+ // Optional step to make less noise on the page, if any!
+ textarea.style.height = 0;
+ // Now append it to your page somewhere, I chose
+ document.body.appendChild(textarea);
+ // Give our textarea a value of whatever inside the div of id=containerid
+ textarea.value = document.getElementById(containerid).innerText;
+ // Now copy whatever inside the textarea to clipboard
+ var selector = document.querySelector('#temp_element');
+ selector.select();
+ document.execCommand('copy');
+ // Remove the textarea
+ document.body.removeChild(textarea);
};
diff --git a/javascript/parsers.js b/javascript/parsers.js
new file mode 100644
index 0000000..0cd165c
--- /dev/null
+++ b/javascript/parsers.js
@@ -0,0 +1,54 @@
+
+function WhisperParser() {
+
+}
+
+WhisperParser.prototype.detect = function (jsonobj) {
+ let retval = (jsonobj.segments) ? true : false
+ if(retval) { console.log('Whisper formatted data detected'); }
+ return retval
+}
+
+/**
+ * configure editor toggles given the transcript input
+ */
+WhisperParser.prototype.config = function () {
+ // turn off confidence toggle
+ document.getElementById('confidence').removeAttribute('disabled');
+}
+
+WhisperParser.prototype.generate = function (jsonobj) {
+ var results = jsonobj.segments
+
+ transcriptObject = results.words;
+ var confidence = 1;
+
+ // process each result
+ results.forEach((segment, idx) => {
+ let adjustment = 0
+
+ let word_start_time = segment.start + adjustment;
+ let word_start_time_ms = word_start_time * 1000
+
+ let duration = segment.end - segment.start
+ let duration_ms = 1000 * duration
+
+ let confidence = Math.exp(segment.avg_logprob) // see: https://github.com/openai/whisper/discussions/1183#discussioncomment-5585981
+ let paraId = "para-" + idx
+ let speakerId = "speaker 0"
+
+ newPara = CreateNewPara(word_start_time, speakerId, paraId);
+ $('#content').append(newPara);
+
+ let html = ""
+ + " "
+ + segment.text
+ + "";
+
+ para = "#" + paraId;
+
+ $(para).append(html);
+
+ })
+}
+