From 9be8a443d74d68f179de88fff13b4e8424579d7b Mon Sep 17 00:00:00 2001
From: zhifu gao <zhifu.gzf@alibaba-inc.com>
Date: 星期五, 10 三月 2023 18:24:39 +0800
Subject: [PATCH] Merge pull request #207 from alibaba-damo-academy/dev_dzh
---
funasr/utils/timestamp_tools.py | 56 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
1 files changed, 56 insertions(+), 0 deletions(-)
diff --git a/funasr/utils/timestamp_tools.py b/funasr/utils/timestamp_tools.py
index f966aee..4a367f8 100644
--- a/funasr/utils/timestamp_tools.py
+++ b/funasr/utils/timestamp_tools.py
@@ -4,7 +4,10 @@
import numpy as np
from typing import Any, List, Tuple, Union
+
def time_stamp_lfr6_pl(us_alphas, us_cif_peak, char_list, begin_time=0.0, end_time=None):
+ if not len(char_list):
+ return []
START_END_THRESHOLD = 5
TIME_RATE = 10.0 * 6 / 1000 / 3 # 3 times upsampled
if len(us_alphas.shape) == 3:
@@ -52,3 +55,56 @@
res.append([int(timestamp[0] * 1000), int(timestamp[1] * 1000)])
return res
+
+def time_stamp_sentence(punc_id_list, time_stamp_postprocessed, text_postprocessed):
+ res = []
+ if text_postprocessed is None:
+ return res
+ if time_stamp_postprocessed is None:
+ return res
+ if len(time_stamp_postprocessed) == 0:
+ return res
+ if len(text_postprocessed) == 0:
+ return res
+ if punc_id_list is None or len(punc_id_list) == 0:
+ res.append({
+ 'text': text_postprocessed.split(),
+ "start": time_stamp_postprocessed[0][0],
+ "end": time_stamp_postprocessed[-1][1]
+ })
+ return res
+ if len(punc_id_list) != len(time_stamp_postprocessed):
+ res.append({
+ 'text': text_postprocessed.split(),
+ "start": time_stamp_postprocessed[0][0],
+ "end": time_stamp_postprocessed[-1][1]
+ })
+ return res
+
+ sentence_text = ''
+ sentence_start = time_stamp_postprocessed[0][0]
+ texts = text_postprocessed.split()
+ for i in range(len(punc_id_list)):
+ sentence_text += texts[i]
+ if punc_id_list[i] == 2:
+ sentence_text += ','
+ res.append({
+ 'text': sentence_text,
+ "start": sentence_start,
+ "end": time_stamp_postprocessed[i][1]
+ })
+ sentence_text = ''
+ sentence_start = time_stamp_postprocessed[i][1]
+ elif punc_id_list[i] == 3:
+ sentence_text += '.'
+ res.append({
+ 'text': sentence_text,
+ "start": sentence_start,
+ "end": time_stamp_postprocessed[i][1]
+ })
+ sentence_text = ''
+ sentence_start = time_stamp_postprocessed[i][1]
+ return res
+
+
+
--
Gitblit v1.9.1