speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

export

Browse Source
This commit is contained in:
游雁 2023-03-29 13:20:27 +08:00
parent ee3f4e5236
commit c039cbc3bf
2 changed files with 28 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
funasr/runtime/python/onnxruntime/demo_vad.py
View File

@ -1,12 +1,30 @@
import soundfile
from funasr_onnx import Fsmn_vad
model_dir = "/Users/zhifu/Downloads/speech_fsmn_vad_zh-cn-16k-common-pytorch"
wav_path = "/Users/zhifu/Downloads/speech_fsmn_vad_zh-cn-16k-common-pytorch/example/vad_example.wav"
model = Fsmn_vad(model_dir)
wav_path = "/Users/zhifu/Downloads/speech_fsmn_vad_zh-cn-16k-common-pytorch/example/vad_example.wav"
#offline vad
# result = model(wav_path)
# print(result)
#online vad
speech, sample_rate = soundfile.read(wav_path)
speech_length = speech.shape[0]
sample_offset = 0
step = 160 * 10
param_dict = {'in_cache': []}
for sample_offset in range(0, speech_length, min(step, speech_length - sample_offset)):
if sample_offset + step >= speech_length - 1:
step = speech_length - sample_offset
is_final = True
else:
is_final = False
param_dict['is_final'] = is_final
segments_result = model(audio_in=speech[sample_offset: sample_offset + step],
param_dict=param_dict)
print(segments_result)
result = model(wav_path)
print(result)

10
funasr/runtime/python/onnxruntime/funasr_onnx/vad_bin.py
View File

@ -53,13 +53,13 @@ class Fsmn_vad():
proj_dim = self.encoder_conf["proj_dim"]
lorder = self.encoder_conf["lorder"]
for i in range(fsmn_layers):
cache = np.zeros(1, proj_dim, lorder-1, 1).astype(np.float32)
cache = np.zeros((1, proj_dim, lorder-1, 1)).astype(np.float32)
in_cache.append(cache)
return in_cache
def __call__(self, wav_content: Union[str, np.ndarray, List[str]], **kwargs) -> List:
waveform_list = self.load_data(wav_content, self.frontend.opts.frame_opts.samp_freq)
def __call__(self, audio_in: Union[str, np.ndarray, List[str]], **kwargs) -> List:
waveform_list = self.load_data(audio_in, self.frontend.opts.frame_opts.samp_freq)
waveform_nums = len(waveform_list)
is_final = kwargs.get('kwargs', False)
@ -70,13 +70,13 @@ class Fsmn_vad():
waveform = waveform_list[beg_idx:end_idx]
feats, feats_len = self.extract_feat(waveform)
param_dict = kwargs.get('param_dict', dict())
in_cache = param_dict.get('cache', list())
in_cache = param_dict.get('in_cache', list())
in_cache = self.prepare_cache(in_cache)
try:
inputs = [feats]
inputs.extend(in_cache)
scores, out_caches = self.infer(inputs)
param_dict['cache'] = out_caches
param_dict['in_cache'] = out_caches
segments = self.vad_scorer(scores, waveform[0][None, :], is_final=is_final, max_end_sil=self.max_end_sil)
except ONNXRuntimeError: