diff --git a/test/unit/agents/transcription/test_speech_recognizer.py b/test/unit/agents/transcription/test_speech_recognizer.py index 88a5ac2..ab28dcf 100644 --- a/test/unit/agents/transcription/test_speech_recognizer.py +++ b/test/unit/agents/transcription/test_speech_recognizer.py @@ -5,12 +5,13 @@ from control_backend.agents.transcription.speech_recognizer import OpenAIWhisper def test_estimate_max_tokens(): - """Inputting one minute of audio, assuming 300 words per minute, expecting 400 tokens.""" + """Inputting one minute of audio, assuming 450 words per minute and adding a 10 token padding, + expecting 610 tokens.""" audio = np.empty(shape=(60 * 16_000), dtype=np.float32) actual = SpeechRecognizer._estimate_max_tokens(audio) - assert actual == 400 + assert actual == 610 assert isinstance(actual, int)