Spaces:
Runtime error
Runtime error
| from transformers import pipeline | |
| import gradio as gr | |
| import whisper | |
| wav2vec_models = { | |
| "en" : pipeline("automatic-speech-recognition", model="facebook/wav2vec2-base-960h"), | |
| "fr" : pipeline("automatic-speech-recognition", model="facebook/wav2vec2-large-xlsr-53-french"), | |
| "es" : pipeline("automatic-speech-recognition", model="facebook/wav2vec2-large-xlsr-53-spanish"), | |
| "it" : pipeline("automatic-speech-recognition", model="facebook/wav2vec2-large-xlsr-53-italian"), | |
| "ru" : pipeline("automatic-speech-recognition", model="anton-l/wav2vec2-large-xlsr-53-russian") | |
| } | |
| whisper_model = whisper.load_model("base") | |
| def transcribe_audio(language=None, mic=None, file=None): | |
| if mic is not None: | |
| audio = mic | |
| elif file is not None: | |
| audio = file | |
| else: | |
| return "You must either provide a mic recording or a file" | |
| wav2vec_model = wav2vec_models[language] | |
| transcription = wav2vec_model(audio)["text"] | |
| transcription2 = whisper_model.transcribe(audio, language=language)["text"] | |
| return transcription, transcription2 | |
| title = "Speech2text comparison (Wav2vec vs Whisper)" | |
| description = """ | |
| This Space allows easy comparisons for transcribed texts between Facebook's Wav2vec model and newly released OpenAI's Whisper model.\n | |
| (Even if Whisper includes a language detection and even an automatic translation, here we have decided to select the language to speed up the transcription and to focus only on the quality of the transcriptions. The default language is english) | |
| """ | |
| article = "Check out [the OpenAI Whisper model](https://github.com/openai/whisper) and [the Facebook Wav2vec model](https://ai.facebook.com/blog/wav2vec-20-learning-the-structure-of-speech-from-raw-audio/) that this demo is based off of." | |
| examples = [["en", None, "english_sentence.flac"], | |
| ["en", None, "6_Steps_To_Hit_ANY_Goal.mp3000.mp3"], | |
| ["fr", None, "2022-a-Droite-un-fauteuil-pour-trois-3034044.mp3000.mp3"], | |
| ["fr", None, "podcast-bdl-episode-5-mix-v2.mp3000.mp3"], | |
| ["es", None, "momiasartesecretodelantiguoegipto-nationalgeographicespana-ivoox73191074.mp3000.mp3"], | |
| ["es", None, "millonarioscohetesrepresentaestanuev-xataka-ivoox73148634.mp3000.mp3"], | |
| ["it", None, "Ansa_voice_barbero_no_sigla.mp3000.mp3"], | |
| ["it", None, "A304176327.mp3000.mp3"], | |
| ["ru", None, "ekonomika-s-nikitoy-krichevskim-21-07-2021-podcast6127-serie444725.mp3000.mp3"]] | |
| gr.Interface( | |
| fn=transcribe_audio, | |
| inputs=[ | |
| gr.Radio(label="Language", choices=["en", "fr", "es","it", "ru"], value="en"), | |
| gr.Audio(source="microphone", type="filepath", optional=True), | |
| gr.Audio(source="upload", type="filepath", optional=True), | |
| ], | |
| outputs=[ | |
| gr.Textbox(label="facebook/wav2vec"), | |
| gr.Textbox(label="openai/whisper"),], | |
| title=title, | |
| description=description, | |
| article=article, | |
| examples=examples | |
| ).launch(debug=True) |