app.py

import spaces
import gradio as gr
# Use a pipeline as a high-level helper
import torch
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline, AutoModelForCausalLM, AutoTokenizer

@spaces.GPU(duration=120)
def transcribe_audio(audio, model_id):
    if audio is None:
        return "Please upload an audio file."
    if model_id is None:
        return "Please select a model."

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

    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=25,
        batch_size=16,
        torch_dtype=torch_dtype,
        device=device,
    )

    result = pipe(audio)
    return result["text"]

@spaces.GPU(duration=180)
def proofread(text):
    if text is None:
        return "Please provide the transcribed text for proofreading."
    
    device = "cuda:0" if torch.cuda.is_available() else "cpu"
    torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
    
    prompt = "用繁體中文整理這段文字，分段及改正錯別字，最後加上整段文字的重點。"
    
    model = AutoModelForCausalLM.from_pretrained("hfl/llama-3-chinese-8b-instruct-v3")
    tokenizer = AutoTokenizer.from_pretrained("hfl/llama-3-chinese-8b-instruct-v3")
    model.to(device)
    
    # Perform proofreading using the model
    input_text = prompt + text
    input_ids = tokenizer.encode(input_text, return_tensors="pt").to(device)
    output = model.generate(input_ids, max_length=len(input_ids[0])+50, num_return_sequences=1, temperature=0.7)
    proofread_text = tokenizer.decode(output[0], skip_special_tokens=True)
    
    return proofread_text


with gr.Blocks() as demo:
    gr.Markdown("""
                # Audio Transcription and Proofreading
                1. Upload an audio file (Wait for the file to be fully loaded first)
                2. Select a model for transcription
                3. Proofread the transcribed text
                """)

    with gr.Row():
        with gr.Column():
            audio = gr.Audio(sources="upload", type="filepath")
            video = gr.Video(sources="upload", type="url")
        model_dropdown = gr.Dropdown(choices=["openai/whisper-large-v3", "alvanlii/whisper-small-cantonese"], value="openai/whisper-large-v3")

    transcribe_button = gr.Button("Transcribe")
    transcribed_text = gr.Textbox(label="Transcribed Text")
    
    proofread_button = gr.Button("Proofread")
    proofread_output = gr.Textbox(label="Proofread Text")

    transcribe_button.click(transcribe_audio, inputs=[audio, model_dropdown], outputs=transcribed_text)
    proofread_button.click(proofread, inputs=[transcribed_text], outputs=proofread_output)
    transcribed_text.change(proofread, inputs=[transcribed_text], outputs=proofread_output)

demo.launch()