app.py

import gradio as gr
import json
import torch
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline

device = "cuda:0" if torch.cuda.is_available() else "cpu"
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32

model_id = "openai/whisper-large-v3"

model = AutoModelForSpeechSeq2Seq.from_pretrained(
    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
)
model.to(device)

processor = AutoProcessor.from_pretrained(model_id)

pipe = pipeline(
    "automatic-speech-recognition",
    model=model,
    tokenizer=processor.tokenizer,
    feature_extractor=processor.feature_extractor,
    max_new_tokens=128,
    chunk_length_s=30,
    batch_size=16,
    return_timestamps=True,
    torch_dtype=torch_dtype,
    device=device,
)


def process_audio(audio_file):
    # In this example, let's just return a hardcoded array of JSON objects
    output_data = [
        {"label": "cat", "confidence": 0.8},
        {"label": "dog", "confidence": 0.7},
        {"label": "bird", "confidence": 0.6}
    ]
    return json.dumps(output_data)


def process(audio):
    # Read audio data from the file
    # with open(audio.name, 'rb') as f:
    #     audio_data = f.read()
    audio_data, audio_filename = audio


    # Process the audio data
    result = pipe(audio_data)['chunks']
    for item in result:
        item['timestamp'] = list(item['timestamp'])
    return json.dumps(result)


iface = gr.Interface(fn=process, inputs="audio", outputs="text")
iface.launch()