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

Merge pull request #293 from alibaba-damo-academy/dev_zly

Browse Source 
support max_end_sil
This commit is contained in:
zhifu gao 2023-03-24 13:46:34 +08:00 committed by GitHub
commit 5c4bd2b718
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 11 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
egs_modelscope/vad/speech_fsmn_vad_zh-cn-16k-common/infer_online.py
View File

@ -22,7 +22,7 @@ if __name__ == '__main__':
sample_offset = 0
step = 160 * 10
param_dict = {'in_cache': dict()}
param_dict = {'in_cache': dict(), 'max_end_sil': 800}
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

2
egs_modelscope/vad/speech_fsmn_vad_zh-cn-8k-common/infer_online.py
View File

@ -22,7 +22,7 @@ if __name__ == '__main__':
sample_offset = 0
step = 80 * 10
param_dict = {'in_cache': dict()}
param_dict = {'in_cache': dict(), 'max_end_sil': 800}
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

10
funasr/bin/vad_inference_online.py
View File

@ -30,7 +30,8 @@ from funasr.models.frontend.wav_frontend import WavFrontendOnline
from funasr.models.frontend.wav_frontend import WavFrontend
from funasr.bin.vad_inference import Speech2VadSegment
header_colors = '\033[95m'
end_colors = '\033[0m'
class Speech2VadSegmentOnline(Speech2VadSegment):
@ -55,7 +56,7 @@ class Speech2VadSegmentOnline(Speech2VadSegment):
@torch.no_grad()
def __call__(
self, speech: Union[torch.Tensor, np.ndarray], speech_lengths: Union[torch.Tensor, np.ndarray] = None,
in_cache: Dict[str, torch.Tensor] = dict(), is_final: bool = False
in_cache: Dict[str, torch.Tensor] = dict(), is_final: bool = False, max_end_sil: int = 800
) -> Tuple[torch.Tensor, List[List[int]], torch.Tensor]:
"""Inference
@ -86,7 +87,8 @@ class Speech2VadSegmentOnline(Speech2VadSegment):
"feats": feats,
"waveform": waveforms,
"in_cache": in_cache,
"is_final": is_final
"is_final": is_final,
"max_end_sil": max_end_sil
}
# a. To device
batch = to_device(batch, device=self.device)
@ -217,6 +219,7 @@ def inference_modelscope(
vad_results = []
batch_in_cache = param_dict['in_cache'] if param_dict is not None else dict()
is_final = param_dict['is_final'] if param_dict is not None else False
max_end_sil = param_dict['max_end_sil'] if param_dict is not None else 800
for keys, batch in loader:
assert isinstance(batch, dict), type(batch)
assert all(isinstance(s, str) for s in keys), keys
@ -224,6 +227,7 @@ def inference_modelscope(
assert len(keys) == _bs, f"{len(keys)} != {_bs}"
batch['in_cache'] = batch_in_cache
batch['is_final'] = is_final
batch['max_end_sil'] = max_end_sil
# do vad segment
_, results, param_dict['in_cache'] = speech2vadsegment(**batch)

3
funasr/models/e2e_vad.py Executable file → Normal file
View File

@ -473,8 +473,9 @@ class E2EVadModel(nn.Module):
return segments, in_cache
def forward_online(self, feats: torch.Tensor, waveform: torch.tensor, in_cache: Dict[str, torch.Tensor] = dict(),
is_final: bool = False
is_final: bool = False, max_end_sil: int = 800
) -> Tuple[List[List[List[int]]], Dict[str, torch.Tensor]]:
self.max_end_sil_frame_cnt_thresh = max_end_sil - self.vad_opts.speech_to_sil_time_thres
self.waveform = waveform # compute decibel for each frame
self.ComputeDecibel()
self.ComputeScores(feats, in_cache)