Merge pull request #729 from herrjemand/herrjemand-oai-whisper-api-fix

Updated to the latest OpenAI API changes, and fixed #720
Uberi · Mar 28, 2024 · 4a5b375 · 4a5b375
2 parents 1b737c5 + 9a694d7
commit 4a5b375
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 3 deletions.
diff --git a/speech_recognition/recognizers/whisper.py b/speech_recognition/recognizers/whisper.py
@@ -38,5 +38,6 @@ def recognize_whisper_api(
     wav_data = BytesIO(audio_data.get_wav_data())
     wav_data.name = "SpeechRecognition_audio.wav"
 
-    transcript = openai.Audio.transcribe(model, wav_data, api_key=api_key)
-    return transcript["text"]
+    client = openai.OpenAI(api_key=api_key)
+    transcript = client.audio.transcriptions.create(file=wav_data, model=model)
+    return transcript.text
diff --git a/tests/test_recognition.py b/tests/test_recognition.py
@@ -84,7 +84,7 @@ def test_ibm_chinese(self):
     def test_whisper_english(self):
         r = sr.Recognizer()
         with sr.AudioFile(self.AUDIO_FILE_EN) as source: audio = r.record(source)
-        self.assertEqual(r.recognize_whisper(audio, language="english", **self.WHISPER_CONFIG), " 1, 2, 3.")
+        self.assertEqual(r.recognize_whisper(audio, language="english", **self.WHISPER_CONFIG), " 1, 2, 3")
 
     def test_whisper_french(self):
         r = sr.Recognizer()