v2.1 Python (#360)

.github/workflows/python-demos.yml

@@ -39,6 +39,12 @@ jobs:
     - name: Pre-build dependencies
       run: python -m pip install --upgrade pip
 
+    # ************** REMOVE AFTER RELEASE ********************
+    - name: Build binding
+      run: |
+        pip install wheel setuptools && cd ../../binding/python && python setup.py sdist bdist_wheel && pip install dist/pvcheetah-2.1.0-py3-none-any.whl
+    # ********************************************************
+
     - name: Install dependencies
       run: pip install -r requirements.txt
 
@@ -55,6 +61,12 @@ jobs:
     steps:
     - uses: actions/checkout@v3
 
+    # ************** REMOVE AFTER RELEASE ********************
+    - name: Build binding
+      run: |
+        pip3 install wheel setuptools && cd ../../binding/python && python3 setup.py sdist bdist_wheel && pip3 install dist/pvcheetah-2.1.0-py3-none-any.whl
+    # ********************************************************
+
     - name: Install dependencies
       run: pip3 install -r requirements.txt
 

binding/python/_cheetah.py

@@ -1,12 +1,12 @@
 #
-#    Copyright 2018-2023 Picovoice Inc.
+# Copyright 2018-2024 Picovoice Inc.
 #
-#    You may not use this file except in compliance with the license. A copy of the license is located in the "LICENSE"
-#    file accompanying this source.
+# You may not use this file except in compliance with the license. A copy of the license is located in the "LICENSE"
+# file accompanying this source.
 #
-#    Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
-#    an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
-#    specific language governing permissions and limitations under the License.
+# Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+# an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+# specific language governing permissions and limitations under the License.
 #
 
 import os
@@ -16,7 +16,7 @@
 
 
 class CheetahError(Exception):
-    def __init__(self, message: str = '', message_stack: Sequence[str] = None):
+    def __init__(self, message: str = '', message_stack: Optional[Sequence[str]] = None):
         super().__init__(message)
 
         self._message = message

binding/python/setup.py

@@ -1,5 +1,5 @@
 #
-# Copyright 2022-2023 Picovoice Inc.
+# Copyright 2022-2024 Picovoice Inc.
 #
 # You may not use this file except in compliance with the license. A copy of the license is located in the "LICENSE"
 # file accompanying this source.
@@ -48,7 +48,7 @@
 
 setuptools.setup(
     name="pvcheetah",
-    version="2.0.2",
+    version="2.1.0",
     author="Picovoice",
     author_email="[email protected]",
     description="Cheetah Speech-to-Text Engine.",

binding/python/test_cheetah.py

@@ -1,19 +1,17 @@
 #
-#    Copyright 2018-2023 Picovoice Inc.
+# Copyright 2018-2024 Picovoice Inc.
 #
-#    You may not use this file except in compliance with the license. A copy of the license is located in the "LICENSE"
-#    file accompanying this source.
+# You may not use this file except in compliance with the license. A copy of the license is located in the "LICENSE"
+# file accompanying this source.
 #
-#    Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
-#    an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
-#    specific language governing permissions and limitations under the License.
+# Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+# an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+# specific language governing permissions and limitations under the License.
 #
 
 import os
-import struct
 import sys
 import unittest
-import wave
 
 from parameterized import parameterized
 
@@ -22,7 +20,7 @@
 from test_util import *
 
 
-parameters = load_test_data()
+language_tests = load_test_data()
 
 
 class CheetahTestCase(unittest.TestCase):
@@ -32,25 +30,25 @@ def setUpClass(cls):
         cls._audio_directory = os.path.join('..', '..', 'resources', 'audio_samples')
 
     @classmethod
-    def _create_cheetah(cls, enable_automatic_punctuation: bool) -> Cheetah:
+    def _create_cheetah(cls, language: str, enable_automatic_punctuation: bool) -> Cheetah:
         return Cheetah(
             access_key=cls._access_key,
-            model_path=default_model_path('../..'),
+            model_path=get_model_path_by_language(language=language),
             library_path=default_library_path('../..'),
             enable_automatic_punctuation=enable_automatic_punctuation)
 
-    @parameterized.expand(parameters)
+    @parameterized.expand(language_tests)
     def test_process(
             self,
-            _: str,
+            language: str,
             audio_file: str,
             expected_transcript: str,
             punctuations: List[str],
             error_rate: float):
         o = None
 
         try:
-            o = self._create_cheetah(False)
+            o = self._create_cheetah(language=language, enable_automatic_punctuation=False)
 
             pcm = read_wav_file(
                 file_name=os.path.join(self._audio_directory, audio_file),
@@ -77,18 +75,18 @@ def test_process(
             if o is not None:
                 o.delete()
 
-    @parameterized.expand(parameters)
+    @parameterized.expand(language_tests)
     def test_process_with_punctuation(
             self,
-            _: str,
+            language: str,
             audio_file: str,
             expected_transcript: str,
             punctuations: List[str],
             error_rate: float):
         o = None
 
         try:
-            o = self._create_cheetah(True)
+            o = self._create_cheetah(language=language, enable_automatic_punctuation=True)
 
             pcm = read_wav_file(
                 file_name=os.path.join(self._audio_directory, audio_file),
@@ -112,7 +110,11 @@ def test_process_with_punctuation(
                 o.delete()
 
     def test_version(self):
-        o = self._create_cheetah(False)
+        o = Cheetah(
+            access_key=self._access_key,
+            model_path=default_model_path('../..'),
+            library_path=default_library_path('../..'),
+            enable_automatic_punctuation=True)
         self.assertIsInstance(o.version, str)
         self.assertGreater(len(o.version), 0)
 

binding/python/test_cheetah_perf.py

@@ -1,12 +1,12 @@
 #
-#    Copyright 2022 Picovoice Inc.
+# Copyright 2022-2024 Picovoice Inc.
 #
-#    You may not use this file except in compliance with the license. A copy of the license is located in the "LICENSE"
-#    file accompanying this source.
+# You may not use this file except in compliance with the license. A copy of the license is located in the "LICENSE"
+# file accompanying this source.
 #
-#    Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
-#    an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
-#    specific language governing permissions and limitations under the License.
+# Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+# an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+# specific language governing permissions and limitations under the License.
 #
 
 import os

binding/python/test_util.py

@@ -1,5 +1,5 @@
 #
-# Copyright 2023 Picovoice Inc.
+# Copyright 2023-2024 Picovoice Inc.
 #
 # You may not use this file except in compliance with the license. A copy of the license is located in the "LICENSE"
 # file accompanying this source.
@@ -9,23 +9,32 @@
 # specific language governing permissions and limitations under the License.
 #
 
+import json
+import os
 import struct
 import wave
+
 from typing import *
 
 
 def load_test_data() -> List[Tuple[str, str, str, List[str], float]]:
-    parameters = [
+    data_file_path = os.path.join(os.path.dirname(__file__), "../../resources/.test/test_data.json")
+    with open(data_file_path, encoding="utf8") as data_file:
+        json_test_data = data_file.read()
+    test_data = json.loads(json_test_data)['tests']
+
+    language_tests = [
         (
-            "en",
-            "test.wav",
-            "Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.",
-            ["."],
-            0.025
+            t['language'],
+            t['audio_file'],
+            t['transcript'],
+            t['punctuations'],
+            t['error_rate'],
         )
+        for t in test_data['language_tests']
     ]
 
-    return parameters
+    return language_tests
 
 
 def read_wav_file(file_name: str, sample_rate: int) -> Tuple:
@@ -48,14 +57,25 @@ def read_wav_file(file_name: str, sample_rate: int) -> Tuple:
     return frames[::channels]
 
 
+def get_model_path_by_language(language: str) -> str:
+    model_path_subdir = _append_language('../../lib/common/cheetah_params', language)
+    return os.path.join(os.path.dirname(__file__), f'{model_path_subdir}.pv')
+
+
+def _append_language(s: str, language: str) -> str:
+    if language == 'en':
+        return s
+    return "%s_%s" % (s, language)
+
+
 def get_word_error_rate(transcript: str, expected_transcript: str, use_cer: bool = False) -> float:
     transcript_split = list(transcript) if use_cer else transcript.split()
     expected_split = list(expected_transcript) if use_cer else expected_transcript.split()
     return _levenshtein_distance(transcript_split, expected_split) / len(transcript)
 
 
 def _levenshtein_distance(words1: Sequence[str], words2: Sequence[str]) -> int:
-    res = [[0] * (len(words1) + 2) for _ in range(len(words2) + 1)]
+    res = [[0] * (len(words2) + 1) for _ in range(len(words1) + 1)]
     for i in range(len(words1) + 1):
         res[i][0] = i
     for j in range(len(words2) + 1):

demo/python/requirements.txt

@@ -1,2 +1,2 @@
-pvcheetah==2.0.2
+pvcheetah==2.1.0
 pvrecorder==1.2.3

demo/python/setup.py

@@ -28,7 +28,7 @@
 
 setuptools.setup(
     name="pvcheetahdemo",
-    version="2.0.2",
+    version="2.1.0",
     author="Picovoice",
     author_email="[email protected]",
     description="Cheetah speech-to-text engine demos",