fix eng word space (#598)

Co-authored-by: mengzhe.cmz <mengzhe.cmz@alibaba-inc.com>
This commit is contained in:
chenmengzheAAA 2023-06-06 19:16:36 +08:00 committed by GitHub
parent 94abecd0d9
commit 12e1849c6d
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -103,9 +103,10 @@ string CTTransformer::AddPunc(const char* sz_input)
vector<string> WordWithPunc;
for (int i = 0; i < InputStr.size(); i++)
{
if (i > 0 && !(InputStr[i][0] & 0x80) && (i + 1) <InputStr.size() && !(InputStr[i+1][0] & 0x80))// <20>м<EFBFBD><D0BC>Ӣ<EFBFBD>ģ<EFBFBD>
#if (i > 0 && !(InputStr[i][0] & 0x80) && (i + 1) <InputStr.size() && !(InputStr[i+1][0] & 0x80))// <20>м<EFBFBD><D0BC>Ӣ<EFBFBD>ģ<EFBFBD>
if (i > 0 && !(InputStr[i-1][0] & 0x80) && !(InputStr[i][0] & 0x80))
{
InputStr[i] = InputStr[i]+ " ";
InputStr[i] = " " + InputStr[i];
}
WordWithPunc.push_back(InputStr[i]);
@ -187,4 +188,4 @@ vector<int> CTTransformer::Infer(vector<int32_t> input_data)
return punction;
}
} // namespace funasr
} // namespace funasr