app.py

import sys

import gradio as gr
import spaces
from huggingface_hub import snapshot_download

sys.path.append("./fireredasr")
from fireredasr.models.fireredasr import FireRedAsr


asr_model_aed = None
asr_model_llm = None


def init_model(model_dir_aed, model_dir_llm):
    global asr_model_aed
    global asr_model_llm
    if asr_model_aed is None:
        asr_model_aed = FireRedAsr.from_pretrained("aed", model_dir_aed)
    if asr_model_llm is None:
        asr_model_llm = FireRedAsr.from_pretrained("llm", model_dir_llm)


@spaces.GPU(duration=20)
def asr_inference(audio_file):
    if not audio_file:
        return "Please upload a wav file"
    batch_uttid = ["demo"]
    batch_wav_path = [audio_file]
    results = asr_model_aed.transcribe(
        batch_uttid,
        batch_wav_path,
        {
            "use_gpu": True,
            "beam_size": 3,
            "nbest": 1,
            "decode_max_len": 0,
            "softmax_smoothing": 1.25,
            "aed_length_penalty": 0.6,
            "eos_penalty": 1.0,
            #"decode_min_len": args.decode_min_len,
            #"repetition_penalty": args.repetition_penalty,
            #"llm_length_penalty": args.llm_length_penalty,
            #"temperature": args.temperature
        }
    )
    text_output = results[0]["text"]
    return text_output


@spaces.GPU(duration=30)
def asr_inference_llm(audio_file):
    if not audio_file:
        return "Please upload a wav file"
    batch_uttid = ["demo"]
    batch_wav_path = [audio_file]
    results = asr_model_llm.transcribe(
        batch_uttid,
        batch_wav_path,
        {
            "use_gpu": True,
            "beam_size": 3,
            "nbest": 1,
            "decode_max_len": 0,
            "decode_min_len": 0,
            "repetition_penalty": 3.0,
            "llm_length_penalty": 1.0,
            "temperature": 1.0
        }
    )
    text_output = results[0]["text"]
    return text_output


with gr.Blocks(title="FireRedASR") as demo:
    gr.HTML(
        "<h1 style='text-align: center'>FireRedASR Demo</h1>"
    )
    gr.Markdown("Upload an audio file (wav) to get speech-to-text results.")

    with gr.Row():
        with gr.Column():
            #audio_file = gr.Audio(label="Upload Audio", sources=["upload", "microphone"], type="filepath")
            audio_file = gr.Audio(label="Upload wav file", sources=["upload"], type="filepath")

        with gr.Column():
            asr_button = gr.Button("Start Recognition (FireRedASR-AED-L)", variant="primary")
            text_output = gr.Textbox(label="Model Result (FireRedASR-AED-L)", interactive=False, lines=3, max_lines=12)
            asr_button_llm = gr.Button("Start Recognition (FireRedASR-LLM-L)", variant="primary")
            text_output_llm = gr.Textbox(label="Model Result (FireRedASR-LLM-L)", interactive=False, lines=3, max_lines=12)

    asr_button.click(
        fn=asr_inference,
        inputs=[audio_file],
        outputs=[text_output]
    )

    asr_button_llm.click(
        fn=asr_inference_llm,
        inputs=[audio_file],
        outputs=[text_output_llm]
    )


if __name__ == "__main__":
    # Download model
    local_dir='pretrained_models/FireRedASR-AED-L'
    snapshot_download(repo_id='FireRedTeam/FireRedASR-AED-L', local_dir=local_dir)
    local_dir_llm='pretrained_models/FireRedASR-LLM-L'
    snapshot_download(repo_id='FireRedTeam/FireRedASR-LLM-L', local_dir=local_dir_llm)
    local_dir_qwen='pretrained_models/FireRedASR-LLM-L/Qwen2-7B-Instruct'
    snapshot_download(repo_id='Qwen/Qwen2-7B-Instruct', local_dir=local_dir_qwen)
    # Init model
    init_model(local_dir, local_dir_llm)
    # UI
    demo.queue()
    demo.launch()