app.py

import os
import re
import tempfile

import torch
import gradio as gr
from transformers import pipeline
from pydub import AudioSegment
from pyannote.audio import Pipeline as DiarizationPipeline
import opencc

import spaces  # zeroGPU support
from funasr import AutoModel
from funasr.utils.postprocess_utils import rich_transcription_postprocess

# —————— Model Lists ——————
WHISPER_MODELS = [
    # Base Whisper models
    "openai/whisper-large-v3-turbo",
    "openai/whisper-large-v3",
    "openai/whisper-medium",
    "openai/whisper-small",
    "openai/whisper-base",
    "openai/whisper-tiny",
    # Community fine-tuned Chinese models
    "JacobLinCool/whisper-large-v3-turbo-common_voice_19_0-zh-TW",
    "Jingmiao/whisper-small-zh_tw",
    "DDTChen/whisper-medium-zh-tw",
    "kimbochen/whisper-small-zh-tw",
    # ...etc...
]

SENSEVOICE_MODELS = [
    "FunAudioLLM/SenseVoiceSmall",
    "AXERA-TECH/SenseVoice",
    "alextomcat/SenseVoiceSmall",
    "ChenChenyu/SenseVoiceSmall-finetuned",
    "apinge/sensevoice-small",
]

# —————— Language Options ——————
WHISPER_LANGUAGES = [
    "auto", "af","am","ar","as","az","ba","be","bg","bn","bo",
    "br","bs","ca","cs","cy","da","de","el","en","es","et",
    "eu","fa","fi","fo","fr","gl","gu","ha","haw","he","hi",
    "hr","ht","hu","hy","id","is","it","ja","jw","ka","kk",
    "km","kn","ko","la","lb","ln","lo","lt","lv","mg","mi",
    "mk","ml","mn","mr","ms","mt","my","ne","nl","nn","no",
    "oc","pa","pl","ps","pt","ro","ru","sa","sd","si","sk",
    "sl","sn","so","sq","sr","su","sv","sw","ta","te","tg",
    "th","tk","tl","tr","tt","uk","ur","uz","vi","yi","yo",
    "zh","yue"
]

SENSEVOICE_LANGUAGES = ["auto", "zh", "yue", "en", "ja", "ko", "nospeech"]

# —————— Caches ——————
whisper_pipes = {}
sense_models = {}
dar_pipe = None

# Initialize OpenCC converter for simplified to traditional Chinese
converter = opencc.OpenCC('s2t.json')

# —————— Helpers ——————
def get_whisper_pipe(model_id: str, device: int):
    key = (model_id, device)
    if key not in whisper_pipes:
        whisper_pipes[key] = pipeline(
            "automatic-speech-recognition",
            model=model_id,
            device=device,
            chunk_length_s=30,
            stride_length_s=5,
            return_timestamps=False,
        )
    return whisper_pipes[key]


def get_sense_model(model_id: str):
    if model_id not in sense_models:
        device_str = "cuda:0" if torch.cuda.is_available() else "cpu"
        sense_models[model_id] = AutoModel(
            model=model_id,
            vad_model="fsmn-vad",
            vad_kwargs={"max_single_segment_time": 300000},
            device=device_str,
            hub="hf",
        )
    return sense_models[model_id]


def get_diarization_pipe():
    global dar_pipe
    if dar_pipe is None:
        # Pull token from environment (HF_TOKEN or HUGGINGFACE_TOKEN)
        token = os.getenv("HF_TOKEN") or os.getenv("HUGGINGFACE_TOKEN")
        # Try loading latest 3.1 pipeline, fallback to 2.1 on gated model error
        try:
            dar_pipe = DiarizationPipeline.from_pretrained(
                "pyannote/speaker-diarization-3.1",
                use_auth_token=token or True
            )
        except Exception as e:
            print(f"Failed to load pyannote/speaker-diarization-3.1: {e}\nFalling back to pyannote/[email protected].")
            dar_pipe = DiarizationPipeline.from_pretrained(
                "pyannote/[email protected]",
                use_auth_token=token or True
            )
    return dar_pipe

# —————— Transcription Functions ——————
def transcribe_whisper(model_id: str,
                       language: str,
                       audio_path: str,
                       device_sel: str,
                       enable_diar: bool):
    # select device: 0 for GPU, -1 for CPU
    use_gpu = (device_sel == "GPU" and torch.cuda.is_available())
    device = 0 if use_gpu else -1
    pipe = get_whisper_pipe(model_id, device)
    # full transcription
    result = (pipe(audio_path) if language == "auto"
              else pipe(audio_path, generate_kwargs={"language": language}))
    transcript = result.get("text", "").strip()
    # convert simplified Chinese to traditional
    transcript = converter.convert(transcript)
    diar_text = ""
    # optional speaker diarization
    if enable_diar:
        diarizer = get_diarization_pipe()
        diary = diarizer(audio_path)
        snippets = []
        for turn, _, speaker in diary.itertracks(yield_label=True):
            start_ms, end_ms = int(turn.start*1000), int(turn.end*1000)
            segment = AudioSegment.from_file(audio_path)[start_ms:end_ms]
            with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp:
                segment.export(tmp.name, format="wav")
                seg_out = (pipe(tmp.name) if language == "auto"
                           else pipe(tmp.name, generate_kwargs={"language": language}))
            os.unlink(tmp.name)
            text = seg_out.get("text", "").strip()
            # convert simplified Chinese to traditional
            text = converter.convert(text)
            snippets.append(f"[{speaker}] {text}")
        diar_text = "\n".join(snippets)
    return transcript, diar_text

@spaces.GPU
def transcribe_sense(model_id: str,
                     language: str,
                     audio_path: str,
                     enable_punct: bool,
                     enable_diar: bool):
    model = get_sense_model(model_id)
    # no diarization
    if not enable_diar:
        segs = model.generate(
            input=audio_path,
            cache={},
            language=language,
            use_itn=True,
            batch_size_s=300,
            merge_vad=True,
            merge_length_s=15,
        )
        text = rich_transcription_postprocess(segs[0]['text'])
        if not enable_punct:
            text = re.sub(r"[^\w\s]", "", text)
        # convert simplified Chinese to traditional
        text = converter.convert(text)
        return text, ""
    # with diarization
    diarizer = get_diarization_pipe()
    diary = diarizer(audio_path)
    snippets = []
    for turn, _, speaker in diary.itertracks(yield_label=True):
        start_ms, end_ms = int(turn.start*1000), int(turn.end*1000)
        segment = AudioSegment.from_file(audio_path)[start_ms:end_ms]
        with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmp:
            segment.export(tmp.name, format="wav")
            segs = model.generate(
                input=tmp.name,
                cache={},
                language=language,
                use_itn=True,
                batch_size_s=300,
                merge_vad=False,
                merge_length_s=0,
            )
        os.unlink(tmp.name)
        txt = rich_transcription_postprocess(segs[0]['text'])
        if not enable_punct:
            txt = re.sub(r"[^\w\s]", "", txt)
        # convert simplified Chinese to traditional
        txt = converter.convert(txt)
        snippets.append(f"[{speaker}] {txt}")
    full = rich_transcription_postprocess(model.generate(
        input=audio_path,
        cache={},
        language=language,
        use_itn=True,
        batch_size_s=300,
        merge_vad=True,
        merge_length_s=15
    )[0]['text'])
    if not enable_punct:
        full = re.sub(r"[^\w\s]", "", full)
    full = converter.convert(full)
    return full, "\n".join(snippets)

# —————— Gradio UI ——————
demo = gr.Blocks()
with demo:
    gr.Markdown("## Whisper vs. SenseVoice (Language, Device & Diarization with Simplified→Traditional Chinese)")

    audio_input = gr.Audio(sources=["upload","microphone"], type="filepath", label="Audio Input")

    with gr.Row():
        with gr.Column():
            gr.Markdown("### Whisper ASR")
            whisper_dd = gr.Dropdown(choices=WHISPER_MODELS, value=WHISPER_MODELS[0], label="Whisper Model")
            whisper_lang = gr.Dropdown(choices=WHISPER_LANGUAGES, value="auto", label="Whisper Language")
            device_radio = gr.Radio(choices=["GPU","CPU"], value="GPU", label="Device")
            diar_check = gr.Checkbox(label="Enable Diarization", value=True)
            btn_w = gr.Button("Transcribe with Whisper")
            out_w = gr.Textbox(label="Transcript")
            out_w_d = gr.Textbox(label="Diarized Transcript")
            btn_w.click(fn=transcribe_whisper,
                        inputs=[whisper_dd, whisper_lang, audio_input, device_radio, diar_check],
                        outputs=[out_w, out_w_d])
        with gr.Column():
            gr.Markdown("### FunASR SenseVoice ASR")
            sense_dd = gr.Dropdown(choices=SENSEVOICE_MODELS, value=SENSEVOICE_MODELS[0], label="SenseVoice Model")
            sense_lang = gr.Dropdown(choices=SENSEVOICE_LANGUAGES, value="auto", label="SenseVoice Language")
            punct_chk = gr.Checkbox(label="Enable Punctuation", value=True)
            diar_s_chk = gr.Checkbox(label="Enable Diarization", value=True)
            btn_s = gr.Button("Transcribe with SenseVoice")
            out_s = gr.Textbox(label="Transcript")
            out_s_d = gr.Textbox(label="Diarized Transcript")
            btn_s.click(fn=transcribe_sense,
                        inputs=[sense_dd, sense_lang, audio_input, punct_chk, diar_s_chk],
                        outputs=[out_s, out_s_d])

if __name__ == "__main__":
    demo.launch()