Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
通过缓存seg_dict,加快seaco_paraformer推理 (#2520)
王梦迪561bdbdfc0f8
diff --git a/funasr/models/seaco_paraformer/model.py b/funasr/models/seaco_paraformer/model.pyindex 3b6595cb8..b30c5cb98 100644--- a/funasr/models/seaco_paraformer/model.py+++ b/funasr/models/seaco_paraformer/model.py@@ -16,6 +16,7 @@ import numpy as npfrom typing import Dict, Tuplefrom contextlib import contextmanagerfrom distutils.version import LooseVersion+from functools import lru_cachefrom funasr.register import tablesfrom funasr.utils import postprocess_utils@@ -510,17 +511,6 @@ class SeacoParaformer(BiCifParaformer, Paraformer):return results, meta_datadef generate_hotwords_list(self, hotword_list_or_file, tokenizer=None, frontend=None):- def load_seg_dict(seg_dict_file):- seg_dict = {}- assert isinstance(seg_dict_file, str)- with open(seg_dict_file, "r", encoding="utf8") as f:- lines = f.readlines()- for line in lines:- s = line.strip().split()- key = s[0]- value = s[1:]- seg_dict[key] = " ".join(value)- return seg_dictdef seg_tokenize(txt, seg_dict):pattern = re.compile(r"^[\u4E00-\u9FA50-9]+$")@@ -626,3 +616,17 @@ class SeacoParaformer(BiCifParaformer, Paraformer):models = export_rebuild_model(model=self, **kwargs)return models+++@lru_cache(maxsize=1)+def load_seg_dict(seg_dict_file):+ seg_dict = {}+ assert isinstance(seg_dict_file, str)+ with open(seg_dict_file, "r", encoding="utf8") as f:+ lines = f.readlines()+ for line in lines:+ s = line.strip().split()+ key = s[0]+ value = s[1:]+ seg_dict[key] = " ".join(value)+ return seg_dict\ No newline at end of file