FunASR java http client

VirtuosoQ

2024-04-26 e9d2cfc3a134b00f4e98271fbee3838d1ccecbcc

 examples/industrial_data_pretraining/sense_voice/demo.py

New file
@@ -0,0 +1,24 @@
#!/usr/bin/env python3
# -*- encoding: utf-8 -*-
# Copyright FunASR (https://github.com/alibaba-damo-academy/FunASR). All Rights Reserved.
#  MIT License  (https://opensource.org/licenses/MIT)

from funasr import AutoModel

model = AutoModel(model="/Users/zhifu/Downloads/modelscope_models/SenseVoiceModelscope",
                  vad_model="iic/speech_fsmn_vad_zh-cn-16k-common-pytorch",
              vad_kwargs={"max_single_segment_time": 30000},
                  )


input_wav = "https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav"

DecodingOptions = {
   "task": ("ASR", "AED", "SER"),
   "language": "auto",
   "fp16": True,
   "gain_event": True,
   }

res = model.generate(input=input_wav, batch_size_s=0, DecodingOptions=DecodingOptions)
print(res)