Spaces:
Runtime error
Runtime error
File size: 2,440 Bytes
9568e5e 1500574 9568e5e 1500574 9568e5e 50710a8 1500574 9568e5e 1500574 9568e5e 966d76f 9568e5e ed97bcc 9568e5e ed97bcc 9568e5e ed97bcc 9568e5e 1500574 966d76f 1500574 8472c6f 1500574 966d76f 1500574 966d76f 9568e5e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 |
import torch
import torchaudio
import gradio as gr
from transformers import Wav2Vec2BertProcessor, Wav2Vec2BertForCTC
# Set device
device = "cuda" if torch.cuda.is_available() else "cpu"
# Load processor & model
model_name = "cdactvm/w2v-bert-punjabi" # Change if using a Punjabi ASR model
processor = Wav2Vec2BertProcessor.from_pretrained(model_name)
model = Wav2Vec2BertForCTC.from_pretrained(model_name, torch_dtype=torch.bfloat16).to(device)
def transcribe(audio_path):
# Load audio file
waveform, sample_rate = torchaudio.load(audio_path)
# Convert stereo to mono (if needed)
if waveform.shape[0] > 1:
waveform = torch.mean(waveform, dim=0, keepdim=True)
# Resample to 16kHz
if sample_rate != 16000:
waveform = torchaudio.transforms.Resample(orig_freq=sample_rate, new_freq=16000)(waveform)
# Process audio
inputs = processor(waveform.squeeze(0), sampling_rate=16000, return_tensors="pt")
inputs = {key: val.to(device, dtype=torch.bfloat16) for key, val in inputs.items()}
# Get logits & transcribe
with torch.no_grad():
logits = model(**inputs).logits
predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)[0]
return transcription
# Gradio Interface
app = gr.Interface(
fn=transcribe,
inputs=gr.Audio(sources="upload", type="filepath"),
outputs="text",
title="Punjabi Speech-to-Text",
description="Upload an audio file and get the transcription in Punjabi."
)
if __name__ == "__main__":
app.launch()
# import gradio as gr
# import torch
# from transformers import pipeline
# # Set device
# device = "cuda" if torch.cuda.is_available() else "cpu"
# # Load ASR pipeline
# asr_pipeline = pipeline(
# "automatic-speech-recognition",
# model="cdactvm/w2v-bert-punjabi", # Replace with a Punjabi ASR model if available
# torch_dtype=torch.bfloat16,
# device=0 if torch.cuda.is_available() else -1 # GPU (0) or CPU (-1)
# )
# def transcribe(audio_path):
# # Run inference
# result = asr_pipeline(audio_path)
# return result["text"]
# # Gradio Interface
# app = gr.Interface(
# fn=transcribe,
# inputs=gr.Audio(sources="upload", type="filepath"),
# outputs="text",
# title="Punjabi Speech-to-Text",
# description="Upload an audio file and get the transcription in Punjabi."
# )
# if __name__ == "__main__":
# app.launch()
|