| import gradio as gr | |
| from transformers import Wav2Vec2ForCTC, AutoProcessor | |
| import torch | |
| import numpy as np | |
| import librosa | |
| model_id = "facebook/mms-1b-all" | |
| def transcribe(audio_file_mic=None, audio_file_upload=None): | |
| if audio_file_mic: | |
| audio_file = audio_file_mic | |
| elif audio_file_upload: | |
| audio_file = audio_file_upload | |
| else: | |
| return "Please upload an audio file or record one" | |
| speech, sample_rate = librosa.load(audio_file) | |
| if sample_rate != 16000: | |
| speech = librosa.resample(speech, orig_sr=sample_rate, target_sr=16000) | |
| processor = AutoProcessor.from_pretrained(model_id) | |
| model = Wav2Vec2ForCTC.from_pretrained(model_id) | |
| inputs = processor(speech, sampling_rate=16_000, return_tensors="pt") | |
| with torch.no_grad(): | |
| outputs = model(**inputs).logits | |
| ids = torch.argmax(outputs, dim=-1)[0] | |
| transcription = processor.decode(ids) | |
| return transcription | |
| iface = gr.Interface(fn=transcribe, | |
| inputs=[ | |
| gr.Audio(source="microphone", type="filepath"), | |
| gr.Audio(source="upload", type="filepath") | |
| ], | |
| outputs=["textbox"], | |
| ) | |
| iface.launch() |