From 648c03fdbba045acf8e25c5e6f7df0f27937c3cf Mon Sep 17 00:00:00 2001
From: liugz18 <57401541+liugz18@users.noreply.github.com>
Date: 星期三, 12 六月 2024 12:30:00 +0800
Subject: [PATCH] Update auto_model.py
---
funasr/auto/auto_model.py | 57 ++++++++++++++++++++++++++++++++++++++++-----------------
1 files changed, 40 insertions(+), 17 deletions(-)
diff --git a/funasr/auto/auto_model.py b/funasr/auto/auto_model.py
index e8ca008..9f8e4dc 100644
--- a/funasr/auto/auto_model.py
+++ b/funasr/auto/auto_model.py
@@ -19,6 +19,7 @@
from funasr.utils.load_utils import load_bytes
from funasr.download.file import download_from_url
from funasr.utils.timestamp_tools import timestamp_sentence
+from funasr.utils.timestamp_tools import timestamp_sentence_en
from funasr.download.download_from_hub import download_model
from funasr.utils.vad_utils import slice_padding_audio_samples
from funasr.utils.vad_utils import merge_vad
@@ -233,6 +234,8 @@
# fp16
if kwargs.get("fp16", False):
model.to(torch.float16)
+ elif kwargs.get("bf16", False):
+ model.to(torch.bfloat16)
return model, kwargs
def __call__(self, *args, **cfg):
@@ -321,7 +324,7 @@
input, input_len=input_len, model=self.vad_model, kwargs=self.vad_kwargs, **cfg
)
end_vad = time.time()
-
+
# FIX(gcf): concat the vad clips for sense vocie model for better aed
if kwargs.get("merge_vad", False):
for i in range(len(res)):
@@ -359,6 +362,7 @@
results_sorted = []
if not len(sorted_data):
+ results_ret_list.append({"key": key, "text": "", "timestamp": []})
logging.info("decoding, utt: {}, empty speech".format(key))
continue
@@ -426,6 +430,10 @@
# f"time_speech_total_per_sample: {time_speech_total_per_sample: 0.3f}, "
# f"time_escape_total_per_sample: {time_escape_total_per_sample:0.3f}")
+ if len(results_sorted) != n:
+ results_ret_list.append({"key": key, "text": "", "timestamp": []})
+ logging.info("decoding, utt: {}, empty result".format(key))
+ continue
restored_data = [0] * n
for j in range(n):
index = sorted_data[j][1]
@@ -461,10 +469,11 @@
return_raw_text = kwargs.get("return_raw_text", False)
# step.3 compute punc model
+ raw_text = None
if self.punc_model is not None:
if not len(result["text"].strip()):
if return_raw_text:
- result["raw_text"] = ""
+ result["raw_text"] = raw_text = ""
else:
deep_update(self.punc_kwargs, cfg)
punc_res = self.inference(
@@ -474,9 +483,7 @@
if return_raw_text:
result["raw_text"] = raw_text
result["text"] = punc_res[0]["text"]
- else:
- raw_text = None
-
+
# speaker embedding cluster after resorted
if self.spk_model is not None and kwargs.get("return_spk_res", True):
if raw_text is None:
@@ -512,24 +519,40 @@
and 'iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch'\
can predict timestamp, and speaker diarization relies on timestamps."
)
- sentence_list = timestamp_sentence(
- punc_res[0]["punc_array"],
- result["timestamp"],
- raw_text,
- return_raw_text=return_raw_text,
- )
+ if kwargs.get("en_post_proc", False):
+ sentence_list = timestamp_sentence_en(
+ punc_res[0]["punc_array"],
+ result["timestamp"],
+ raw_text,
+ return_raw_text=return_raw_text,
+ )
+ else:
+ sentence_list = timestamp_sentence(
+ punc_res[0]["punc_array"],
+ result["timestamp"],
+ raw_text,
+ return_raw_text=return_raw_text,
+ )
distribute_spk(sentence_list, sv_output)
result["sentence_info"] = sentence_list
elif kwargs.get("sentence_timestamp", False):
if not len(result["text"].strip()):
sentence_list = []
else:
- sentence_list = timestamp_sentence(
- punc_res[0]["punc_array"],
- result["timestamp"],
- raw_text,
- return_raw_text=return_raw_text,
- )
+ if kwargs.get("en_post_proc", False):
+ sentence_list = timestamp_sentence_en(
+ punc_res[0]["punc_array"],
+ result["timestamp"],
+ raw_text,
+ return_raw_text=return_raw_text,
+ )
+ else:
+ sentence_list = timestamp_sentence(
+ punc_res[0]["punc_array"],
+ result["timestamp"],
+ raw_text,
+ return_raw_text=return_raw_text,
+ )
result["sentence_info"] = sentence_list
if "spk_embedding" in result:
del result["spk_embedding"]
--
Gitblit v1.9.1