import torch import gradio as gr import yt_dlp as youtube_dl from transformers import pipeline from transformers.pipelines.audio_utils import ffmpeg_read import tempfile import os MODEL_NAME = "openai/whisper-large-v3" BATCH_SIZE = 8 FILE_LIMIT_MB = 100000 YT_LENGTH_LIMIT_S = 360000 # limit to 1 hour YouTube files device = 0 if torch.cuda.is_available() else "cpu" pipe = pipeline( task="automatic-speech-recognition", model=MODEL_NAME, chunk_length_s=30, device=device, ) all_special_ids = pipe.tokenizer.all_special_ids transcribe_token_id = all_special_ids[-5] translate_token_id = all_special_ids[-6] def transcribe(microphone, file_upload, task): warn_output = "" if (microphone is not None) and (file_upload is not None): warn_output = ( "警告：您已经上传了一个音频文件并使用了麦克录制。 " "录制文件将被使用上传的音频将被丢弃。" ) elif (microphone is None) and (file_upload is None): return "错误: 您必须使用麦克风录制或上传音频文件" file = microphone if microphone is not None else file_upload pipe.model.config.forced_decoder_ids = [ [2, transcribe_token_id if task == "transcribe" else translate_token_id] ] # text = pipe(file, return_timestamps=True)["text"] text = pipe(file, return_timestamps=True) # trans to SRT text = convert_to_srt(text) return warn_output + text def _return_yt_html_embed(yt_url): video_id = yt_url.split("?v=")[-1] HTML_str = ( f'
' "
" ) return HTML_str def download_yt_audio(yt_url, filename): info_loader = youtube_dl.YoutubeDL() try: info = info_loader.extract_info(yt_url, download=False) except youtube_dl.utils.DownloadError as err: raise gr.Error(str(err)) file_length = info["duration_string"] file_h_m_s = file_length.split(":") file_h_m_s = [int(sub_length) for sub_length in file_h_m_s] if len(file_h_m_s) == 1: file_h_m_s.insert(0, 0) if len(file_h_m_s) == 2: file_h_m_s.insert(0, 0) file_length_s = file_h_m_s[0] * 3600 + file_h_m_s[1] * 60 + file_h_m_s[2] if file_length_s > YT_LENGTH_LIMIT_S: yt_length_limit_hms = time.strftime("%HH:%MM:%SS", time.gmtime(YT_LENGTH_LIMIT_S)) file_length_hms = time.strftime("%HH:%MM:%SS", time.gmtime(file_length_s)) raise gr.Error(f"Maximum YouTube length is {yt_length_limit_hms}, got {file_length_hms} YouTube video.") ydl_opts = {"outtmpl": filename, "format": "worstvideo[ext=mp4]+bestaudio[ext=m4a]/best[ext=mp4]/best"} with youtube_dl.YoutubeDL(ydl_opts) as ydl: try: ydl.download([yt_url]) except youtube_dl.utils.ExtractorError as err: raise gr.Error(str(err)) def yt_transcribe(yt_url, task): html_embed_str = _return_yt_html_embed(yt_url) with tempfile.TemporaryDirectory() as tmpdirname: filepath = os.path.join(tmpdirname, "video.mp4") download_yt_audio(yt_url, filepath) with open(filepath, "rb") as f: inputs = f.read() inputs = ffmpeg_read(inputs, pipe.feature_extractor.sampling_rate) inputs = {"array": inputs, "sampling_rate": pipe.feature_extractor.sampling_rate} text = pipe(inputs,return_timestamps=True) # text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"language":"zh"}, return_timestamps=True) # text = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps=True)["text"] # text = pipe("audio.mp3",return_timestamps=True) #trans to SRT text= convert_to_srt(text) return html_embed_str, text # SRT prepare # Assuming srt format is a sequence of subtitles with index, time range and text def convert_to_srt(input): output = "" index = 1 for chunk in input["chunks"]: start, end = chunk["timestamp"] text = chunk["text"] if end is None: end = "None" # Convert seconds to hours:minutes:seconds,milliseconds format start = format_time(start) end = format_time(end) output += f"{index}\n{start} --> {end}\n{text}\n\n" index += 1 return output # Helper function to format time def format_time(seconds): if seconds == "None": return seconds hours = int(seconds // 3600) minutes = int((seconds % 3600) // 60) seconds = int(seconds % 60) milliseconds = int((seconds % 1) * 1000) return f"{hours:02}:{minutes:02}:{seconds:02},{milliseconds:03}" demo = gr.Blocks() mf_transcribe = gr.Interface( fn=transcribe, inputs=[ gr.inputs.Audio(source="microphone", type="filepath", optional=True), gr.inputs.Audio(source="upload", type="filepath", optional=True), gr.inputs.Radio(["transcribe", "translate"], label="Task", default="transcribe"), ], outputs="text", layout="horizontal", theme="huggingface", title="Audio-to-Text-SRT 自动生成字幕", description=( "直接在网页录音或上传音频文件，加入Youtube连接，轻松转换为文字和字幕格式! 本演示采用" f" 模型 [{MODEL_NAME}](https://huggingface.co./{MODEL_NAME}) 和 🤗 Transformers 转换任意长度的" "音视频文件!使用GPU转换效率会大幅提高，大约每小时 $0.6 约相当于人民币 5 元。 如果您有较长内容,需要更快的转换速度，请私信作者微信 1259388，并备注“语音转文字”" ), allow_flagging="never", ) yt_transcribe = gr.Interface( fn=yt_transcribe, inputs=[ gr.inputs.Textbox(lines=1, placeholder="Paste the URL to a YouTube video here", label="YouTube URL"), # gr.inputs.Radio(["转译", "翻译"], label="Task", default="transcribe") gr.inputs.Radio(["transcribe", "translate"], label="Task", default="transcribe"), ], outputs=["html", "text"], layout="horizontal", theme="huggingface", title="Audio-to-Text-SRT 自动生成字幕", description=( "直接在网页录音或上传音频文件，加入Youtube连接，轻松转换为文字和字幕格式! 本演示采用" f" 模型 [{MODEL_NAME}](https://huggingface.co./{MODEL_NAME}) 和 🤗 Transformers 转换任意长度的" "音视频文件!使用GPU转换效率会大幅提高，大约每小时 $0.6 约相当于人民币 5 元。 如果您有较长内容,需要更快的转换速度，请私信作者微信 1259388，并备注“语音转文字”" ), allow_flagging="never", ) # # Load the images # image1 = Image("wechatqrcode.jpg") # image2 = Image("paypalqrcode.png") # # Define a function that returns the images and captions # def display_images(): # return image1, "WeChat Pay", image2, "PayPal" with demo: gr.TabbedInterface([mf_transcribe, yt_transcribe], ["转译音频成文字", "YouTube转字幕"]) # Create a gradio interface with no inputs and four outputs # gr.Interface(display_images, [], [gr.outputs.Image(), gr.outputs.Textbox(), gr.outputs.Image(), gr.outputs.Textbox()], layout="horizontal").launch() demo.launch(enable_queue=True)