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.
## ❓ Questions and Help 我们在语音识别 中,发现经常会丢失最后一个字,这里提供一个测试的demo,麻烦帮忙看下: 环境:Linux ubuntu 在终端运行命令: `funasr ++model="paraformer-zh" ++input=aaaa.wav` 拿到的结果是: [{'key': 'rand_key_2yW4Acq9GFz6Y', 'text': '我 要 打', 'timestamp': [[1830, 2050], [2050, 2270], [2270, 3175]]}] 实际上期待的文字是: 我要打卡,丢失了最后的“卡”字 测试音频见附件,解压后即可使用...
如下图所示,使用speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch进行语音识别,配合使用speech_eres2net_sv_zh-cn_16k-common作为说话人识别,加载speech_eres2net_sv_zh-cn_16k-common时报错了,是还不支持吗?能否支持啊? 
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) ## ❓ Questions and Help 使用容器镜像启动funASR服务后使用客户端无法调用成功,提示vad_handle is null 后台服务启动日志如下: root@ea758f46b6b5:/workspace/FunASR/runtime# bash run_server_2pass.sh \ > --download-model-dir...
#### 用funasr加载SenseVoiceSmall模型提示 not registered #### Code ``` from funasr import AutoModel model = AutoModel(model=p.model) ``` p.model是我传入的参数,是我已经下载好的SenseVoiceSmall模型的路径 然后提示我:AssertionError: C: \REAPER\UserPlugins\funasr \SenseVoiceSmall is not registered - OS: Win10 - FunASR Version: 1.1.4
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) ## ❓ Questions and Help 中文离线转写服务(CPU)如何配置说话人识别模块, 目前没有看到说话人识别的model配置? #### Code nohup bash run_server.sh \ --download-model-dir /workspace/models...
希望能解决以下问题,针对funasr 4.5-cpu 的docker包 1,支持不了并发,只要服务中有未完成任务,这个时候再发送数据,后台服务就直接死掉了重启了 2,当发送文件无效,服务处理出错时,也应该给客户端返回处理错误信息,便于客户端判断处理
我看文档是这么描述的 # damo/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-onnx(时间戳) # damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404-onnx(nn热词) 如果我需要时间戳和热词,那怎么办,是实现不了吗,我理解这里是只能二选一。另外nn热词是什么意思啊
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 用示例代码测试时报错TypeError: expected Tensor as element 1 in argument 0, but got str...
from funasr import AutoModel import logging logging.getLogger().setLevel(logging.INFO) ser_kwargs = {"use_itn": False, "language":"zh", "batch_size":64, "model":"iic/SenseVoiceSmall"} funasr_model = AutoModel(model="paraformer-zh", vad_model="fsmn-vad", punc_model="ct-punc", spk_model="cam++", ser_model="SenseVoiceSmall", ser_kwargs=ser_kwargs, disable_update=True) input_audio ="/Users/whs/PycharmProjects/speakerLog/test.wav" device ="cpu" res = funasr_model.inference_with_vad(input_audio)...
同时我单独使用vad推理,利用sounddevice从麦克风读取数据,用以下 ``` sample_rate = 16000 # 采样率 channels = 1 # 单声道 dtype = "int16" # 数据类型 blocksize = 1024 # 块大小 def record_audio(): with sd.InputStream( samplerate=sample_rate, channels=channels, dtype=dtype, blocksize=blocksize,...