ailia-speech 1.3.0.3__tar.gz → 1.3.0.5__tar.gz

{ailia_speech-1.3.0.3 → ailia_speech-1.3.0.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ailia_speech
-Version: 1.3.0.3
+Version: 1.3.0.5
 Summary: ailia AI Speech
 Home-page: https://ailia.jp/
 Author: ax Inc.
@@ -41,7 +41,6 @@ pip3 install ./
 ## Usage
 
 ```python
-import ailia
 import ailia_speech
 
 import librosa
@@ -50,19 +49,20 @@ import os
 import urllib.request
 
 # Load target audio
-ref_file_path = "demo.wav"
-if not os.path.exists(ref_file_path):
+input_file_path = "demo.wav"
+if not os.path.exists(input_file_path):
 	urllib.request.urlretrieve(
 		"https://github.com/axinc-ai/ailia-models/raw/refs/heads/master/audio_processing/whisper/demo.wa",
 		"demo.wav"
 	)
-audio_waveform, sampling_rate = librosa.load(ref_file_path, mono=True)
+audio_waveform, sampling_rate = librosa.load(input_file_path, mono=True)
 
 # Infer
 speech = ailia_speech.Whisper()
 speech.initialize_model(model_path = "./models/", model_type = ailia_speech.AILIA_SPEECH_MODEL_TYPE_WHISPER_MULTILINGUAL_SMALL)
 recognized_text = speech.transcribe(audio_waveform, sampling_rate)
-print(recognized_text)
+for text in recognized_text:
+	print(text)
 ```
 
 ## API specification

{ailia_speech-1.3.0.3 → ailia_speech-1.3.0.5}/README.md

@@ -28,7 +28,6 @@ pip3 install ./
 ## Usage
 
 ```python
-import ailia
 import ailia_speech
 
 import librosa
@@ -37,19 +36,20 @@ import os
 import urllib.request
 
 # Load target audio
-ref_file_path = "demo.wav"
-if not os.path.exists(ref_file_path):
+input_file_path = "demo.wav"
+if not os.path.exists(input_file_path):
 	urllib.request.urlretrieve(
 		"https://github.com/axinc-ai/ailia-models/raw/refs/heads/master/audio_processing/whisper/demo.wa",
 		"demo.wav"
 	)
-audio_waveform, sampling_rate = librosa.load(ref_file_path, mono=True)
+audio_waveform, sampling_rate = librosa.load(input_file_path, mono=True)
 
 # Infer
 speech = ailia_speech.Whisper()
 speech.initialize_model(model_path = "./models/", model_type = ailia_speech.AILIA_SPEECH_MODEL_TYPE_WHISPER_MULTILINGUAL_SMALL)
 recognized_text = speech.transcribe(audio_waveform, sampling_rate)
-print(recognized_text)
+for text in recognized_text:
+	print(text)
 ```
 
 ## API specification

{ailia_speech-1.3.0.3 → ailia_speech-1.3.0.5}/ailia_speech/__init__.py

@@ -475,19 +475,24 @@ class Whisper(AiliaSpeechModel):
 
         self._check(dll.ailiaSpeechPushInputData(self._instance, audio_waveform, channels, audio_waveform.shape[0] // channels, sampling_rate))
         self._check(dll.ailiaSpeechFinalizeInputData(self._instance))
-        self._check(dll.ailiaSpeechTranscribe(self._instance))
 
-        count = ctypes.c_uint(0)
-        self._check(dll.ailiaSpeechGetTextCount(self._instance, ctypes.byref(count)))
-        results = []
-        for i in range(count.value):
-            text = AILIASpeechText()
-            self._check(dll.ailiaSpeechGetText(self._instance, ctypes.byref(text), AILIA_SPEECH_TEXT_VERSION, i))
-            results.append({"text" : text.text.decode(), "time_stamp_begin" : text.time_stamp_begin, "time_stamp_end" : text.time_stamp_end, "person_id" : text.person_id, "language" : text.language.decode(), "confidence" : text.confidence})
+        while True:
+            complete = ctypes.c_uint(0)
+            self._check(dll.ailiaSpeechComplete(self._instance, ctypes.byref(complete)))
+            if complete.value == 1:
+                break
 
-        self._check(dll.ailiaSpeechResetTranscribeState(self._instance))
+            self._check(dll.ailiaSpeechTranscribe(self._instance))
+
+            count = ctypes.c_uint(0)
+            self._check(dll.ailiaSpeechGetTextCount(self._instance, ctypes.byref(count)))
+            results = []
+            for i in range(count.value):
+                text = AILIASpeechText()
+                self._check(dll.ailiaSpeechGetText(self._instance, ctypes.byref(text), AILIA_SPEECH_TEXT_VERSION, i))
+                yield {"text" : text.text.decode(), "time_stamp_begin" : text.time_stamp_begin, "time_stamp_end" : text.time_stamp_end, "person_id" : text.person_id, "language" : text.language.decode(), "confidence" : text.confidence}
 
-        return results
+        self._check(dll.ailiaSpeechResetTranscribeState(self._instance))
 
     def __del__(self):
         if self._instance:

{ailia_speech-1.3.0.3 → ailia_speech-1.3.0.5}/ailia_speech.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ailia_speech
-Version: 1.3.0.3
+Version: 1.3.0.5
 Summary: ailia AI Speech
 Home-page: https://ailia.jp/
 Author: ax Inc.
@@ -41,7 +41,6 @@ pip3 install ./
 ## Usage
 
 ```python
-import ailia
 import ailia_speech
 
 import librosa
@@ -50,19 +49,20 @@ import os
 import urllib.request
 
 # Load target audio
-ref_file_path = "demo.wav"
-if not os.path.exists(ref_file_path):
+input_file_path = "demo.wav"
+if not os.path.exists(input_file_path):
 	urllib.request.urlretrieve(
 		"https://github.com/axinc-ai/ailia-models/raw/refs/heads/master/audio_processing/whisper/demo.wa",
 		"demo.wav"
 	)
-audio_waveform, sampling_rate = librosa.load(ref_file_path, mono=True)
+audio_waveform, sampling_rate = librosa.load(input_file_path, mono=True)
 
 # Infer
 speech = ailia_speech.Whisper()
 speech.initialize_model(model_path = "./models/", model_type = ailia_speech.AILIA_SPEECH_MODEL_TYPE_WHISPER_MULTILINGUAL_SMALL)
 recognized_text = speech.transcribe(audio_waveform, sampling_rate)
-print(recognized_text)
+for text in recognized_text:
+	print(text)
 ```
 
 ## API specification

{ailia_speech-1.3.0.3 → ailia_speech-1.3.0.5}/setup.py

@@ -54,7 +54,7 @@ if __name__ == "__main__":
     setup(
         name="ailia_speech",
         scripts=scripts,
-        version="1.3.0.3",
+        version="1.3.0.5",
         install_requires=[
             "ailia",
             "ailia_tokenizer",