Merge pull request #163 from alibaba-damo-academy/dev_zly

zhifu gao

2023-02-28 be7230fd94477fdc22a9b315af7c5e3ac4a7eb8d

Merge pull request #163 from alibaba-damo-academy/dev_zly

update vad inference

 funasr/bin/vad_inference.py

@@ -86,7 +86,8 @@

    @torch.no_grad()
    def __call__(
            self, speech: Union[torch.Tensor, np.ndarray], speech_lengths: Union[torch.Tensor, np.ndarray] = None
            self, speech: Union[torch.Tensor, np.ndarray], speech_lengths: Union[torch.Tensor, np.ndarray] = None,
            in_cache: Dict[str, torch.Tensor] = dict()
    ) -> Tuple[List[List[int]], Dict[str, torch.Tensor]]:
        """Inference

@@ -125,11 +126,12 @@
            batch = {
                "feats": feats[:, t_offset:t_offset + step, :],
                "waveform": speech[:, t_offset * 160:min(speech.shape[-1], (t_offset + step - 1) * 160 + 400)],
                "is_final": is_final
                "is_final": is_final,
                "in_cache": in_cache
            }
            # a. To device
            batch = to_device(batch, device=self.device)
            segments_part = self.vad_model(**batch)
            segments_part, in_cache = self.vad_model(**batch)
            if segments_part:
                for batch_num in range(0, self.batch_size):
                    segments[batch_num] += segments_part[batch_num]

 funasr/models/e2e_vad.py

@@ -444,7 +444,7 @@

    def forward(self, feats: torch.Tensor, waveform: torch.tensor, in_cache: Dict[str, torch.Tensor] = dict(),
                is_final: bool = False
                ) -> List[List[List[int]]]:
                ) -> Tuple[List[List[List[int]]], Dict[str, torch.Tensor]]:
        self.waveform = waveform  # compute decibel for each frame
        self.ComputeDecibel()
        self.ComputeScores(feats, in_cache)
@@ -468,8 +468,7 @@
        if is_final:
            # reset class variables and clear the dict for the next query
            self.AllResetDetection()
            in_cache.clear()
        return segments
        return segments, in_cache

    def DetectCommonFrames(self) -> int:
        if self.vad_state_machine == VadStateMachine.kVadInStateEndPointDetected: