| | |
|---|
| | | Deep-FSMN for Large Vocabulary Continuous Speech Recognition |
|---|
| | | https://arxiv.org/abs/1803.05030 |
|---|
| | | """ |
|---|
| | | |
|---|
| | | def __init__( |
|---|
| | | self, |
|---|
| | | sample_rate: int = 16000, |
|---|
| | |
|---|
| | | Deep-FSMN for Large Vocabulary Continuous Speech Recognition |
|---|
| | | https://arxiv.org/abs/1803.05030 |
|---|
| | | """ |
|---|
| | | |
|---|
| | | def __init__(self): |
|---|
| | | self.start_ms = 0 |
|---|
| | | self.end_ms = 0 |
|---|
| | |
|---|
| | | Deep-FSMN for Large Vocabulary Continuous Speech Recognition |
|---|
| | | https://arxiv.org/abs/1803.05030 |
|---|
| | | """ |
|---|
| | | |
|---|
| | | def __init__(self): |
|---|
| | | self.noise_prob = 0.0 |
|---|
| | | self.speech_prob = 0.0 |
|---|
| | |
|---|
| | | Deep-FSMN for Large Vocabulary Continuous Speech Recognition |
|---|
| | | https://arxiv.org/abs/1803.05030 |
|---|
| | | """ |
|---|
| | | |
|---|
| | | def __init__(self, window_size_ms: int, sil_to_speech_time: int, |
|---|
| | | speech_to_sil_time: int, frame_size_ms: int): |
|---|
| | | self.window_size_ms = window_size_ms |
|---|
| | |
|---|
| | | Deep-FSMN for Large Vocabulary Continuous Speech Recognition |
|---|
| | | https://arxiv.org/abs/1803.05030 |
|---|
| | | """ |
|---|
| | | |
|---|
| | | def __init__(self, encoder: FSMN, vad_post_args: Dict[str, Any], frontend=None): |
|---|
| | | super(E2EVadModel, self).__init__() |
|---|
| | | self.vad_opts = VADXOptions(**vad_post_args) |
|---|
| | |
|---|
| | | self.vad_opts.frame_in_ms) |
|---|
| | | self.encoder = encoder |
|---|
| | | # init variables |
|---|
| | | self.is_final = False |
|---|
| | | self.data_buf_start_frame = 0 |
|---|
| | | self.frm_cnt = 0 |
|---|
| | | self.latest_confirmed_speech_frame = 0 |
|---|
| | |
|---|
| | | self.frontend = frontend |
|---|
| | | |
|---|
| | | def AllResetDetection(self): |
|---|
| | | self.is_final = False |
|---|
| | | self.data_buf_start_frame = 0 |
|---|
| | | self.frm_cnt = 0 |
|---|
| | | self.latest_confirmed_speech_frame = 0 |
|---|
| | |
|---|
| | | def forward(self, feats: torch.Tensor, waveform: torch.tensor, in_cache: Dict[str, torch.Tensor] = dict(), |
|---|
| | | is_final: bool = False |
|---|
| | | ) -> Tuple[List[List[List[int]]], Dict[str, torch.Tensor]]: |
|---|
| | | if not in_cache: |
|---|
| | | self.AllResetDetection() |
|---|
| | | self.waveform = waveform # compute decibel for each frame |
|---|
| | | self.ComputeDecibel() |
|---|
| | | self.ComputeScores(feats, in_cache) |
|---|
| | |
|---|
| | | segment_batch = [] |
|---|
| | | if len(self.output_data_buf) > 0: |
|---|
| | | for i in range(self.output_data_buf_offset, len(self.output_data_buf)): |
|---|
| | | if not is_final and ( |
|---|
| | | not self.output_data_buf[i].contain_seg_start_point or not self.output_data_buf[ |
|---|
| | | if not is_final and (not self.output_data_buf[i].contain_seg_start_point or not self.output_data_buf[ |
|---|
| | | i].contain_seg_end_point): |
|---|
| | | continue |
|---|
| | | segment = [self.output_data_buf[i].start_ms, self.output_data_buf[i].end_ms] |
|---|
| | |
|---|
| | | def forward_online(self, feats: torch.Tensor, waveform: torch.tensor, in_cache: Dict[str, torch.Tensor] = dict(), |
|---|
| | | is_final: bool = False, max_end_sil: int = 800 |
|---|
| | | ) -> Tuple[List[List[List[int]]], Dict[str, torch.Tensor]]: |
|---|
| | | if not in_cache: |
|---|
| | | self.AllResetDetection() |
|---|
| | | 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 |
|---|
| | | |
|---|
