Merge pull request #1 from nschloe/readme-fixes

larger images

.github/workflows/ci.yml

@@ -31,6 +31,8 @@ jobs:
     - uses: actions/checkout@v2
       with:
         lfs: true
+    - name: Install system dependencies
+      run: sudo apt install ffmpeg
     - name: Test with tox
       run: |
         pip install tox

README.md

@@ -1,4 +1,7 @@
-# specky
+<p align="center">
+  <a href="https://github.com/nschloe/specky"><img alt="specky" src="https://nschloe.github.io/specky/specky-logo.svg" width="50%"></a>
+  <p align="center">Delicious audio file spectrograms.</p>
+</p>
 
 [![gh-actions](https://img.shields.io/github/workflow/status/nschloe/specky/ci?style=flat-square)](https://github.com/nschloe/specky/actions?query=workflow%3Aci)
 [![Code style: black](https://img.shields.io/badge/code%20style-black-000000.svg?style=flat-square)](https://github.com/psf/black)
@@ -8,10 +11,14 @@
 [![PyPi downloads](https://img.shields.io/pypi/dm/specky.svg?style=flat-square)](https://pypistats.org/packages/specky)
 
 specky is a collection of easy-to-use Python and command-line tools for analyzing audio
-files.
+files. Install from [PyPi](https://pypi.org/project/specky/) with
+```
+pip install specky
+```
+and use with
 ```bash
-specky-show filename.mp3   # shows the wave spectrum of the audio file
-specky-check filename.mp3  # checks if the audio file is worse than it pretends to be
+specky-show filename.mp3                # shows the wave spectrum of the audio file
+specky-check {dirname or filename.mp3}  # checks if the audio file is worse than it pretends to be
 ```
 See `-h` for their respective command-line arguments.
 
@@ -24,13 +31,13 @@ Here is the `specky-show` output of a [sample
 file](https://nschloe.github.io/specky/Yamaha-V50-Ride-Pattern-120bpm.wav) (only first
 channel shown):
 
-<img src="https://nschloe.github.io/specky/wav.png" width="70%"> | <img src="https://nschloe.github.io/specky/320.png" width="70%"> | <img src="https://nschloe.github.io/specky/256.png" width="70%">
+<img src="https://nschloe.github.io/specky/wav.png" width="100%"> | <img src="https://nschloe.github.io/specky/320.png" width="100%"> | <img src="https://nschloe.github.io/specky/256.png" width="100%">
 :-------------------:|:------------------:|:--------------:|
 Full-quality WAV     |  320 kbps MP3      |  256 kbps MP3  |
-<img src="https://nschloe.github.io/specky/192.png" width="70%"> | <img src="https://nschloe.github.io/specky/128.png" width="70%"> | <img src="https://nschloe.github.io/specky/64.png" width="70%">
-|  192 kbps MP3      |  128 kbps MP3  |  64 kbps MP3  |
-<img src="https://nschloe.github.io/specky/96.png" width="70%"> | <img src="https://nschloe.github.io/specky/64.png" width="70%"> | <img src="https://nschloe.github.io/specky/32.png" width="70%">
-|  96 kbps MP3      |  64 kbps MP3  |  32 kbps MP3  |
+<img src="https://nschloe.github.io/specky/192.png" width="100%"> | <img src="https://nschloe.github.io/specky/128.png" width="100%"> | <img src="https://nschloe.github.io/specky/96.png" width="100%">
+|  192 kbps MP3      |  128 kbps MP3  |  96 kbps MP3  |
+<img src="https://nschloe.github.io/specky/64.png" width="100%"> | <img src="https://nschloe.github.io/specky/32.png" width="100%"> | <img src="https://nschloe.github.io/specky/16.png" width="100%">
+|  64 kbps MP3      |  32 kbps MP3  |  16 kbps MP3  |
 
 
 ### License

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = specky
-version = 0.1.0
+version = 0.1.1
 author = Nico Schlömer
 email = [email protected]
 description = Audio file spectral analysis
@@ -31,6 +31,7 @@ install_requires =
     colorama
     importlib_metadata
     matplotlib
+    mutagen
     numpy
     pydub
     scipy

specky/_main.py

@@ -1,3 +1,4 @@
+import glob
 import os
 
 import matplotlib.colors as colors
@@ -9,7 +10,14 @@
 from scipy import signal
 
 
-def show(filename, min_freq=1.0e-2, window_length_s=0.05, channel=None, outfile=None):
+def show(
+    filename,
+    min_freq=1.0e-2,
+    num_windows=None,
+    num_frequencies=None,
+    channel=None,
+    outfile=None,
+):
     track = AudioSegment.from_file(filename)
 
     out = numpy.array(track.get_array_of_samples()).reshape(-1, track.channels)
@@ -19,25 +27,28 @@ def show(filename, min_freq=1.0e-2, window_length_s=0.05, channel=None, outfile=
     else:
         channels = [channel - 1]
 
-    if window_length_s is None:
+    if num_windows is None:
         nperseg = None
     else:
-        nperseg = int(round(window_length_s * track.frame_rate))
-
-    # overlap = 0.5
-    # noverlap = int(round(overlap * window_length_samples))
+        nperseg = int(round(track.duration_seconds / num_windows * track.frame_rate))
 
     for i, k in enumerate(channels):
+        # Perhaps one can downsample this.
+        # https://stackoverflow.com/q/60866162/353337
         f, t, Sxx = signal.spectrogram(
             out[:, k],
             fs=track.frame_rate,
             scaling="spectrum",
             mode="magnitude",
-            # nperseg=window_length_samples
             nperseg=nperseg,
-            # noverlap=noverlap
         )
 
+        if num_frequencies is not None:
+            # ditch some of the frequencies
+            f_step = -(-f.shape[0] // num_frequencies)
+            f = f[::f_step]
+            Sxx = Sxx[::f_step]
+
         # Make sure all values are positive for the log scaling
         smallest_positive = numpy.min(Sxx[Sxx > 0])
         Sxx[Sxx < smallest_positive] = smallest_positive
@@ -59,7 +70,18 @@ def show(filename, min_freq=1.0e-2, window_length_s=0.05, channel=None, outfile=
         plt.savefig(outfile, transparent=True, bbox_inches="tight")
 
 
-def check(filename, min_freq=1.0e-2, window_length_s=0.05, channel=0):
+def check(path, **kwargs):
+    if os.path.isfile(path):
+        _check_file(path, **kwargs)
+        return
+
+    assert os.path.isdir(path)
+    for filename in glob.glob(path + "/**/*.mp3", recursive=True):
+        _check_file(filename, **kwargs)
+    return
+
+
+def _check_file(filename, min_freq=1.0e-2, window_length_s=0.05, channel=0):
     track = AudioSegment.from_file(filename)
 
     out = numpy.array(track.get_array_of_samples()).reshape(-1, track.channels)
@@ -117,7 +139,9 @@ def check(filename, min_freq=1.0e-2, window_length_s=0.05, channel=0):
             expected_max_freq = val
 
         if f[k] > expected_max_freq:
-            print(f"{Fore.GREEN}{filename} seems good.{Style.RESET_ALL}")
+            print(
+                f"{Fore.GREEN}{filename} seems good [{bitrate} kbps].{Style.RESET_ALL}"
+            )
         else:
             print(
                 f"{Fore.RED}{filename} is MP3 [{bitrate} kbps], but has max frequency "

specky/cli/check.py

@@ -8,7 +8,7 @@ def check(argv=None):
     # Parse command line arguments.
     parser = _get_parser()
     args = parser.parse_args(argv)
-    main_check(args.filename)
+    main_check(args.path)
 
 
 def _get_parser():
@@ -19,7 +19,7 @@ def _get_parser():
         formatter_class=argparse.RawTextHelpFormatter,
     )
 
-    parser.add_argument("filename", type=str, help="audio file to analyze")
+    parser.add_argument("path", type=str, help="audio file or directory to analyze")
 
     version_text = "\n".join(
         [

specky/cli/show.py

@@ -8,7 +8,13 @@ def show(argv=None):
     # Parse command line arguments.
     parser = _get_parser()
     args = parser.parse_args(argv)
-    main_show(args.filename, channel=args.channel, outfile=args.outfile)
+    main_show(
+        args.filename,
+        channel=args.channel,
+        outfile=args.outfile,
+        num_windows=args.num_windows,
+        num_frequencies=args.num_frequencies,
+    )
 
 
 def _get_parser():
@@ -29,6 +35,22 @@ def _get_parser():
         help="only show one particular channel (default: show all channels)",
     )
 
+    parser.add_argument(
+        "-w",
+        "--num-windows",
+        default=300,
+        type=int,
+        help="number of windows (x-axis resolution, default: 300)",
+    )
+
+    parser.add_argument(
+        "-f",
+        "--num-frequencies",
+        default=300,
+        type=int,
+        help="number of frequencies (y-axis resolution, default: 300)",
+    )
+
     parser.add_argument(
         "-o",
         "--outfile",