From d20c030e5b75306dd67e8fe9924d5d94eac1bf30 Mon Sep 17 00:00:00 2001
From: wusong <63332221+wusong1128@users.noreply.github.com>
Date: 星期三, 25 九月 2024 15:11:50 +0800
Subject: [PATCH] 解决python ws服务针对尾部非人声录音无结束标识返回的问题 (#2102)
---
funasr/models/whisper/model.py | 8 +++++++-
1 files changed, 7 insertions(+), 1 deletions(-)
diff --git a/funasr/models/whisper/model.py b/funasr/models/whisper/model.py
index 8e9245a..791fddd 100644
--- a/funasr/models/whisper/model.py
+++ b/funasr/models/whisper/model.py
@@ -7,7 +7,11 @@
import torch.nn.functional as F
from torch import Tensor
from torch import nn
+
import whisper
+
+# import whisper_timestamped as whisper
+
from funasr.utils.load_utils import load_audio_text_image_video, extract_fbank
from funasr.register import tables
@@ -108,7 +112,9 @@
# decode the audio
options = whisper.DecodingOptions(**kwargs.get("DecodingOptions", {}))
- result = whisper.decode(self.model, speech, options)
+
+ result = whisper.decode(self.model, speech, options=options)
+ # result = whisper.transcribe(self.model, speech)
results = []
result_i = {"key": key[0], "text": result.text}
--
Gitblit v1.9.1