From 297fafd674715bcd849557616bd22b6791460f31 Mon Sep 17 00:00:00 2001
From: haoneng.lhn <haoneng.lhn@alibaba-inc.com>
Date: 星期四, 04 五月 2023 16:40:34 +0800
Subject: [PATCH] update streaming paraformer text process
---
funasr/bin/asr_inference_paraformer_streaming.py | 13 +++++++------
1 files changed, 7 insertions(+), 6 deletions(-)
diff --git a/funasr/bin/asr_inference_paraformer_streaming.py b/funasr/bin/asr_inference_paraformer_streaming.py
index bf5590c..341abe6 100644
--- a/funasr/bin/asr_inference_paraformer_streaming.py
+++ b/funasr/bin/asr_inference_paraformer_streaming.py
@@ -239,7 +239,7 @@
feats_len = torch.tensor([feats_chunk2.shape[1]])
results_chunk2 = self.infer(feats_chunk2, feats_len, cache)
- return ["".join(results_chunk1 + results_chunk2)]
+ return [" ".join(results_chunk1 + results_chunk2)]
results = self.infer(feats, feats_len, cache)
@@ -299,12 +299,13 @@
# Change integer-ids to tokens
token = self.converter.ids2tokens(token_int)
+ token = " ".join(token)
- if self.tokenizer is not None:
- text = self.tokenizer.tokens2text(token)
- else:
- text = None
- results.append(text)
+ #if self.tokenizer is not None:
+ # text = self.tokenizer.tokens2text(token)
+ #else:
+ # text = None
+ results.append(token)
# assert check_return_type(results)
return results
--
Gitblit v1.9.1