speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Funasr1.0 (#1279)

Browse Source 
* funasr1.0 update

* funasr1.0 paraformer-en
This commit is contained in:
zhifu gao 2024-01-22 15:42:45 +08:00 committed by GitHub
parent 1159adbca0
commit 3919d7454c
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 23 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
.gitignore vendored
View File

@ -23,3 +23,4 @@ samples
.ipynb_checkpoints
outputs*
emotion2vec*
GPT-SoVITS*

6
funasr/models/paraformer/model.py
View File

@ -528,9 +528,9 @@ class Paraformer(torch.nn.Module):
if tokenizer is not None:
# Change integer-ids to tokens
token = tokenizer.ids2tokens(token_int)
text = tokenizer.tokens2text(token)
text_postprocessed, _ = postprocess_utils.sentence_postprocess(token)
text_postprocessed = tokenizer.tokens2text(token)
if not hasattr(tokenizer, "bpemodel"):
text_postprocessed, _ = postprocess_utils.sentence_postprocess(token)
result_i = {"key": key[i], "text": text_postprocessed}

26
funasr/tokenizer/sentencepiece_tokenizer.py
View File

@ -5,12 +5,16 @@ from typing import Union
import sentencepiece as spm
from funasr.tokenizer.abs_tokenizer import AbsTokenizer
from funasr.tokenizer.abs_tokenizer import BaseTokenizer
from funasr.register import tables
class SentencepiecesTokenizer(AbsTokenizer):
def __init__(self, model: Union[Path, str]):
self.model = str(model)
@tables.register("tokenizer_classes", "SentencepiecesTokenizer")
class SentencepiecesTokenizer(BaseTokenizer):
def __init__(self, bpemodel: Union[Path, str],
**kwargs
):
super().__init__(**kwargs)
self.bpemodel = str(bpemodel)
# NOTE(kamo):
# Don't build SentencePieceProcessor in __init__()
# because it's not picklable and it may cause following error,
@ -19,13 +23,13 @@ class SentencepiecesTokenizer(AbsTokenizer):
self.sp = None
def __repr__(self):
return f'{self.__class__.__name__}(model="{self.model}")'
return f'{self.__class__.__name__}(model="{self.bpemodel}")'
def _build_sentence_piece_processor(self):
# Build SentencePieceProcessor lazily.
if self.sp is None:
self.sp = spm.SentencePieceProcessor()
self.sp.load(self.model)
self.sp.load(self.bpemodel)
def text2tokens(self, line: str) -> List[str]:
self._build_sentence_piece_processor()
@ -34,3 +38,11 @@ class SentencepiecesTokenizer(AbsTokenizer):
def tokens2text(self, tokens: Iterable[str]) -> str:
self._build_sentence_piece_processor()
return self.sp.DecodePieces(list(tokens))
def encode(self, line: str) -> List[int]:
self._build_sentence_piece_processor()
return self.sp.EncodeAsIds(line)
def decode(self, line: List[int]):
self._build_sentence_piece_processor()
return self.sp.DecodeIds(line)