From a030ff0f85fd6b1cc2a1d443d2fcfb11ccb1aa8f Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期三, 29 三月 2023 21:15:55 +0800
Subject: [PATCH] export
---
funasr/runtime/python/onnxruntime/funasr_onnx/vad_bin.py | 21 ++++++++++++---------
1 files changed, 12 insertions(+), 9 deletions(-)
diff --git a/funasr/runtime/python/onnxruntime/funasr_onnx/vad_bin.py b/funasr/runtime/python/onnxruntime/funasr_onnx/vad_bin.py
index 533b4b7..cdd4578 100644
--- a/funasr/runtime/python/onnxruntime/funasr_onnx/vad_bin.py
+++ b/funasr/runtime/python/onnxruntime/funasr_onnx/vad_bin.py
@@ -23,7 +23,7 @@
device_id: Union[str, int] = "-1",
quantize: bool = False,
intra_op_num_threads: int = 4,
- max_end_sil: int = 800,
+ max_end_sil: int = None,
):
if not Path(model_dir).exists():
@@ -43,20 +43,23 @@
self.ort_infer = OrtInferSession(model_file, device_id, intra_op_num_threads=intra_op_num_threads)
self.batch_size = batch_size
self.vad_scorer = E2EVadModel(config["vad_post_conf"])
- self.max_end_sil = max_end_sil
+ self.max_end_sil = max_end_sil if max_end_sil is not None else config["vad_post_conf"]["max_end_silence_time"]
+ self.encoder_conf = config["encoder_conf"]
def prepare_cache(self, in_cache: list = []):
if len(in_cache) > 0:
return in_cache
-
- for i in range(4):
- cache = np.random.rand(1, 128, 19, 1).astype(np.float32)
+ fsmn_layers = self.encoder_conf["fsmn_layers"]
+ 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)
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)
@@ -67,13 +70,13 @@
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:
--
Gitblit v1.9.1