From cf8e000a84e888495dcf30c4dbfecea1ee7ab4e2 Mon Sep 17 00:00:00 2001
From: jmwang66 <wangjiaming.wjm@alibaba-inc.com>
Date: 星期一, 07 八月 2023 16:13:37 +0800
Subject: [PATCH] Merge pull request #807 from alibaba-damo-academy/dev_wjm
---
funasr/bin/asr_inference_launch.py | 18 ++++++++----------
1 files changed, 8 insertions(+), 10 deletions(-)
diff --git a/funasr/bin/asr_inference_launch.py b/funasr/bin/asr_inference_launch.py
index 701379a..0fd5abd 100644
--- a/funasr/bin/asr_inference_launch.py
+++ b/funasr/bin/asr_inference_launch.py
@@ -260,8 +260,6 @@
hotword_list_or_file = None
clas_scale = 1.0
- if kwargs.get("device", None) == "cpu":
- ngpu = 0
if ngpu >= 1 and torch.cuda.is_available():
device = "cuda"
else:
@@ -370,7 +368,7 @@
results = speech2text(**batch)
if len(results) < 1:
hyp = Hypothesis(score=0.0, scores={}, states={}, yseq=[])
- results = [[" ", ["sil"], [2], hyp, 10, 6]] * nbest
+ results = [[" ", ["sil"], [2], hyp, 10, 6, []]] * nbest
time_end = time.time()
forward_time = time_end - time_beg
lfr_factor = results[0][-1]
@@ -439,6 +437,7 @@
logging.info(rtf_avg)
if writer is not None:
ibest_writer["rtf"]["rtf_avf"] = rtf_avg
+ torch.cuda.empty_cache()
return asr_result_list
return _forward
@@ -564,6 +563,8 @@
if 'hotword' in kwargs:
hotword_list_or_file = kwargs['hotword']
+ speech2vadsegment.vad_model.vad_opts.max_single_segment_time = kwargs.get("max_single_segment_time", 60000)
+ batch_size_token_threshold_s = kwargs.get("batch_size_token_threshold_s", int(speech2vadsegment.vad_model.vad_opts.max_single_segment_time*0.67/1000)) * 1000
batch_size_token = kwargs.get("batch_size_token", 6000)
print("batch_size_token: ", batch_size_token)
@@ -646,8 +647,7 @@
beg_idx = 0
for j, _ in enumerate(range(0, n)):
batch_size_token_ms_cum += (sorted_data[j][0][1] - sorted_data[j][0][0])
- if j < n - 1 and (batch_size_token_ms_cum + sorted_data[j + 1][0][1] - sorted_data[j + 1][0][
- 0]) < batch_size_token_ms:
+ if j < n - 1 and (batch_size_token_ms_cum + sorted_data[j + 1][0][1] - sorted_data[j + 1][0][0]) < batch_size_token_ms and (sorted_data[j + 1][0][1] - sorted_data[j + 1][0][0]) < batch_size_token_threshold_s:
continue
batch_size_token_ms_cum = 0
end_idx = j + 1
@@ -730,6 +730,7 @@
ibest_writer["time_stamp"][key] = "{}".format(time_stamp_postprocessed)
logging.info("decoding, utt: {}, predictions: {}".format(key, text_postprocessed_punc))
+ torch.cuda.empty_cache()
return asr_result_list
return _forward
@@ -1338,7 +1339,7 @@
format="%(asctime)s (%(module)s:%(lineno)d) %(levelname)s: %(message)s",
)
- if ngpu >= 1:
+ if ngpu >= 1 and torch.cuda.is_available():
device = "cuda"
else:
device = "cpu"
@@ -1369,10 +1370,7 @@
left_context=left_context,
right_context=right_context,
)
- speech2text = Speech2TextTransducer.from_pretrained(
- model_tag=model_tag,
- **speech2text_kwargs,
- )
+ speech2text = Speech2TextTransducer(**speech2text_kwargs)
def _forward(data_path_and_name_and_type,
raw_inputs: Union[np.ndarray, torch.Tensor] = None,
--
Gitblit v1.9.1