我从这里安装了aadnk的whisper-webui, 并已经可以正常的运行了
但是碰到了几个问题:
问题一: 我想修改config.json5配置中vad_cpu_cores的值, 这似乎能加快解析音频的速度, 但是只要改成大于2的值, 就会产生报错
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "L:\AI\Python310\lib\multiprocessing\spawn.py", line 116, in spawn_main
exitcode = _main(fd, parent_sentinel)
File "L:\AI\Python310\lib\multiprocessing\spawn.py", line 125, in _main
prepare(preparation_data)
File "L:\AI\Python310\lib\multiprocessing\spawn.py", line 236, in prepare
_fixup_main_from_path(data['init_main_from_path'])
File "L:\AI\Python310\lib\multiprocessing\spawn.py", line 287, in _fixup_main_from_path
main_content = runpy.run_path(main_path,
File "L:\AI\Python310\lib\runpy.py", line 289, in run_path
return _run_module_code(code, init_globals, run_name,
File "L:\AI\Python310\lib\runpy.py", line 96, in _run_module_code
_run_code(code, mod_globals, init_globals,
File "L:\AI\Python310\lib\runpy.py", line 86, in _run_code
exec(code, run_globals)
File "L:\声音\whisper-webui-main\app-local.py", line 5, in <module>
create_ui(ApplicationConfig.create_default(input_audio_max_duration=-1))
File "L:\声音\whisper-webui-main\app.py", line 714, in create_ui
demo.launch(share=app_config.share, server_name=app_config.server_name, server_port=app_config.server_port)
File "L:\AI\Python310\lib\site-packages\gradio\blocks.py", line 1840, in launch
) = networking.start_server(
File "L:\AI\Python310\lib\site-packages\gradio\networking.py", line 169, in start_server
raise OSError(
OSError: Cannot find empty port in range: 7860-7860. You can specify a different port by setting the GRADIO_SERVER_PORT environment variable or passing the `server_port` parameter to `launch()`.
从报错来看, vad_cpu_cores大于2, 好像会创建多个GRADIO, 导致占用的同样的端口
问题二: 转录较长音频的时候, 虽然后台还是在转录, 但是前端已经报错了, 虽然最终的转录的srt还是能在AppData\Local\Temp\gradio中找到, 但这情况也不对吧
配置文件
{
"models": [
// Configuration for the built-in models. You can remove any of these
// if you don't want to use the default models.
{
"name": "tiny",
"url": "tiny"
},
{
"name": "base",
"url": "base"
},
{
"name": "small",
"url": "small"
},
{
"name": "medium",
"url": "medium"
},
{
"name": "large",
"url": "large"
},
{
"name": "large-v2",
"url": "large-v2"
},
{
"name": "large-v3",
"url": "large-v3"
},
// Uncomment to add custom Japanese models
//{
// "name": "whisper-large-v2-mix-jp",
// "url": "vumichien/whisper-large-v2-mix-jp",
// // The type of the model. Can be "huggingface" or "whisper" - "whisper" is the default.
// // HuggingFace models are loaded using the HuggingFace transformers library and then converted to Whisper models.
// "type": "huggingface",
//},
//{
// "name": "local-model",
// "url": "path/to/local/model",
//},
//{
// "name": "remote-model",
// "url": "https://example.com/path/to/model",
//}
],
// Configuration options that will be used if they are not specified in the command line arguments.
// * WEBUI options *
// Maximum audio file length in seconds, or -1 for no limit. Ignored by CLI.
"input_audio_max_duration": -1,
// True to share the app on HuggingFace.
"share": false,
// The host or IP to bind to. If None, bind to localhost.
"server_name": null,
// The port to bind to.
"server_port": 7860,
// The number of workers to use for the web server. Use -1 to disable queueing.
"queue_concurrency_count": 1,
// Whether or not to automatically delete all uploaded files, to save disk space
"delete_uploaded_files": true,
// * General options *
// The default implementation to use for Whisper. Can be "whisper" or "faster-whisper".
// Note that you must either install the requirements for faster-whisper (requirements-fasterWhisper.txt)
// or whisper (requirements.txt)
"whisper_implementation": "faster-whisper",
// The default model name.
"default_model_name": "medium",
// The default VAD.
"default_vad": "silero-vad",
// A commma delimited list of CUDA devices to use for parallel processing. If None, disable parallel processing.
"vad_parallel_devices": "",
// The number of CPU cores to use for VAD pre-processing.
"vad_cpu_cores": 1,
// The number of seconds before inactivate processes are terminated. Use 0 to close processes immediately, or None for no timeout.
"vad_process_timeout": 1800,
// True to use all available GPUs and CPU cores for processing. Use vad_cpu_cores/vad_parallel_devices to specify the number of CPU cores/GPUs to use.
"auto_parallel": false,
// Directory to save the outputs (CLI will use the current directory if not specified)
"output_dir": null,
// The path to save model files; uses ~/.cache/whisper by default
"model_dir": null,
// Device to use for PyTorch inference, or Null to use the default device
"device": null,
// Whether to print out the progress and debug messages
"verbose": true,
// Whether to perform X->X speech recognition ('transcribe') or X->English translation ('translate')
"task": "transcribe",
// Language spoken in the audio, specify None to perform language detection
"language": "Chinese",
// The window size (in seconds) to merge voice segments
"vad_merge_window": 5,
// The maximum size (in seconds) of a voice segment
"vad_max_merge_size": 30,
// The padding (in seconds) to add to each voice segment
"vad_padding": 1,
// Whether or not to prepend the initial prompt to each VAD segment (prepend_all_segments), or just the first segment (prepend_first_segment)
"vad_initial_prompt_mode": "prepend_first_segment",
// The window size of the prompt to pass to Whisper
"vad_prompt_window": 3,
// Temperature to use for sampling
"temperature": 0,
// Number of candidates when sampling with non-zero temperature
"best_of": 5,
// Number of beams in beam search, only applicable when temperature is zero
"beam_size": 5,
// Optional patience value to use in beam decoding, as in https://arxiv.org/abs/2204.05424, the default (1.0) is equivalent to conventional beam search
"patience": 1,
// Optional token length penalty coefficient (alpha) as in https://arxiv.org/abs/1609.08144, uses simple length normalization by default
"length_penalty": null,
// Comma-separated list of token ids to suppress during sampling; '-1' will suppress most special characters except common punctuations
"suppress_tokens": "-1",
// Optional text to provide as a prompt for the first window
"initial_prompt": null,
// If True, provide the previous output of the model as a prompt for the next window; disabling may make the text inconsistent across windows, but the model becomes less prone to getting stuck in a failure loop
"condition_on_previous_text": true,
// Whether to perform inference in fp16; True by default
"fp16": true,
// The compute type used by faster-whisper. Can be "int8". "int16" or "float16".
"compute_type": "int8",
// Temperature to increase when falling back when the decoding fails to meet either of the thresholds below
"temperature_increment_on_fallback": 0.2,
// If the gzip compression ratio is higher than this value, treat the decoding as failed
"compression_ratio_threshold": 2.4,
// If the average log probability is lower than this value, treat the decoding as failed
"logprob_threshold": -1.0,
// If the probability of the <no-speech> token is higher than this value AND the decoding has failed due to `logprob_threshold`, consider the segment as silence
"no_speech_threshold": 0.6,
// (experimental) extract word-level timestamps and refine the results based on them
"word_timestamps": false,
// if word_timestamps is True, merge these punctuation symbols with the next word
"prepend_punctuations": "\"\'“¿([{-",
// if word_timestamps is True, merge these punctuation symbols with the previous word
"append_punctuations": "\"\'.。,，!！?？:：”)]}、",
// (requires --word_timestamps True) underline each word as it is spoken in srt and vtt
"highlight_words": false,
// Diarization settings
"auth_token": null,
// Whether to perform speaker diarization
"diarization": false,
// The number of speakers to detect
"diarization_speakers": 2,
// The minimum number of speakers to detect
"diarization_min_speakers": 1,
// The maximum number of speakers to detect
"diarization_max_speakers": 8,
// The number of seconds before inactivate processes are terminated. Use 0 to close processes immediately, or None for no timeout.
"diarization_process_timeout": 60,
}
那我是的使用方法错了吗? 还是哪里出了问题
有没有人教教我