FunASR
FunASR copied to clipboard
A Fundamental End-to-End Speech Recognition Toolkit and Open Source SOTA Pretrained Models, Supporting Speech Recognition, Voice Activity Detection, Text Post-processing etc.
数组越界问题
Notice: In order to resolve issues more efficiently, please raise issue following the template. (注意:为了更加高效率解决您遇到的问题,请按照模板提问,补充细节) ## 🐛 Bug ### To Reproduce Steps to reproduce the behavior (**always include the command...
## ❓ Questions and Help ### Before asking: 1. search the issues. 2. search the docs. #### What is your question? #### Code #### What have you tried? #### What's...
您好 我在windows下更新了新的版本(1.0.7)之后,部署的cpu版本的runtime/websocket, online模式,当客户端的话筒输入一段较长时长的连续语音时(10s以上),服务端的funasr-wss-server-2pass.exe程序内存占用会缓慢增长,请问这个问题你们有遇到过吗? #### What's your environment? - OS (e.g., win10,vs2019): - FunASR Version (e.g., 1.0.7): -
## 🐛 Bug When running the asr inference for a blank audio file, there might be a runtime error as shown in the following attached. Take the uploaded audio test...
What is your question? 我用进程方式启动AutoModel处理,16K单声道的wav音频数据,vad模型内部处理数据直接卡住不动,请大佬帮我看看进程启动下vad模型内部处理数据为什么会卡住。 funasr->utils->load_utils.py的64行 data_or_path_or_list = data_or_path_or_list.mean(0) ps:用线程模式就能正常执行,但咱们线程模式长时间运行,有严重的内存泄漏,而且vad不支持多线程。 Code ..... pool = multiprocessing.Pool(2) pool.apply(func=convert, args=(wavepath)) ..... def convert(wavepath): model = AutoModel(model="iic/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch", model_revision="v2.0.4", vad_model="iic/speech_fsmn_vad_zh-cn-16k-common-pytorch", vad_model_revision="v2.0.4", punc_model="iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch", punc_model_revision="v2.0.4", spk_model="iic/speech_campplus_sv_zh-cn_16k-common", spk_model_revision="v2.0.2",...
## 🐛 Bug 和whisperx使用的时候遇到一个版本依赖的问题,如果需要使用最新版funasr和whisperx会出现下列问题 ## 复现流程 1. 如果安转顺序是先安装whisperx,funasr,modelscope默认的whisperx版本是3.1.2,会让whisperx降级,现在whisperx最新版是3.1.3,修复了align使用的bug 2. 如果安装顺序变成先安装funasr,modelscope后安装whisperx会出现datasets版本为2.19.0,这时候下载模型会报错需要降级为2.18.0 ## 正常安装流程 所以要正常安装顺序使用的话就要,先安装funasr,modelscope,whisperx,datasets==2.18.0
在librispeech上测试model = AutoModel(model="paraformer-zh-streaming"),效果不理想
我线下用最新的模型训练好替换官方的模型,但docker启动后仍然下载已经写死版本的模型并覆盖了我的模型,请问下这个能支持自定义吗? 在 FunASR/runtime/websocket/bin/funasr-wss-server.cpp 和 FunASR/runtime/websocket/bin/funasr-wss-server-2pass.cpp 中模型指定版本均已写死,如下所示。 `TCLAP::ValueArg model_revision( "", "model-revision", "ASR model revision", false, "v1.2.1", "string");`
## ❓ Questions and Help 内存溢出,当部署完在线识别服务器之后,进行语音识别(启用热词的情况下),内存占用增长特快, #### What's your environment? - OS (e.g., Linux): - FunASR Version (e.g., 1.0.9): - ModelScope Version (e.g., 1.11.0): - PyTorch Version (e.g., 2.0.0): -...