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
tutorial
游雁3b580e4a7480
diff --git a/docs/tutorial/README_zh.md b/docs/tutorial/README_zh.mdindex 572eef1c5..0e3846df3 100644--- a/docs/tutorial/README_zh.md+++ b/docs/tutorial/README_zh.md@@ -85,6 +85,7 @@ funasr/bin/train.py \++train_conf.validate_interval=2000 \++train_conf.save_checkpoint_interval=2000 \++train_conf.keep_nbest_models=20 \+++train_conf.avg_nbest_model=5 \++optim_conf.lr=0.0002 \++output_dir="${output_dir}" &> ${log_file}```@@ -101,6 +102,7 @@ funasr/bin/train.py \- `train_conf.validate_interval`(int):训练中做验证测试的间隔step数。- `train_conf.save_checkpoint_interval`(int):训练中模型保存间隔step数。- `train_conf.keep_nbest_models`(int):保留最大多少个模型参数,按照验证集acc排序,从高到底保留。+- `train_conf.avg_nbest_model`(int):对acc最高的n个模型取平均。- `optim_conf.lr`(float):学习率。- `output_dir`(str):模型保存路径。- `**kwargs`(dict): 所有在`config.yaml`中参数,均可以直接在此处指定,例如,过滤20s以上长音频:`dataset_conf.max_token_length=2000`,单位为音频fbank帧数(1帧10ms)或者文字token个数。diff --git a/examples/README_zh.md b/examples/README_zh.mdindex 572eef1c5..0e3846df3 100644--- a/examples/README_zh.md+++ b/examples/README_zh.md@@ -85,6 +85,7 @@ funasr/bin/train.py \++train_conf.validate_interval=2000 \++train_conf.save_checkpoint_interval=2000 \++train_conf.keep_nbest_models=20 \+++train_conf.avg_nbest_model=5 \++optim_conf.lr=0.0002 \++output_dir="${output_dir}" &> ${log_file}```@@ -101,6 +102,7 @@ funasr/bin/train.py \- `train_conf.validate_interval`(int):训练中做验证测试的间隔step数。- `train_conf.save_checkpoint_interval`(int):训练中模型保存间隔step数。- `train_conf.keep_nbest_models`(int):保留最大多少个模型参数,按照验证集acc排序,从高到底保留。+- `train_conf.avg_nbest_model`(int):对acc最高的n个模型取平均。- `optim_conf.lr`(float):学习率。- `output_dir`(str):模型保存路径。- `**kwargs`(dict): 所有在`config.yaml`中参数,均可以直接在此处指定,例如,过滤20s以上长音频:`dataset_conf.max_token_length=2000`,单位为音频fbank帧数(1帧10ms)或者文字token个数。diff --git a/examples/industrial_data_pretraining/paraformer/README_zh.md b/examples/industrial_data_pretraining/paraformer/README_zh.mdindex 572eef1c5..0e3846df3 100644--- a/examples/industrial_data_pretraining/paraformer/README_zh.md+++ b/examples/industrial_data_pretraining/paraformer/README_zh.md@@ -85,6 +85,7 @@ funasr/bin/train.py \++train_conf.validate_interval=2000 \++train_conf.save_checkpoint_interval=2000 \++train_conf.keep_nbest_models=20 \+++train_conf.avg_nbest_model=5 \++optim_conf.lr=0.0002 \++output_dir="${output_dir}" &> ${log_file}```@@ -101,6 +102,7 @@ funasr/bin/train.py \- `train_conf.validate_interval`(int):训练中做验证测试的间隔step数。- `train_conf.save_checkpoint_interval`(int):训练中模型保存间隔step数。- `train_conf.keep_nbest_models`(int):保留最大多少个模型参数,按照验证集acc排序,从高到底保留。+- `train_conf.avg_nbest_model`(int):对acc最高的n个模型取平均。- `optim_conf.lr`(float):学习率。- `output_dir`(str):模型保存路径。- `**kwargs`(dict): 所有在`config.yaml`中参数,均可以直接在此处指定,例如,过滤20s以上长音频:`dataset_conf.max_token_length=2000`,单位为音频fbank帧数(1帧10ms)或者文字token个数。diff --git a/examples/industrial_data_pretraining/paraformer/finetune.sh b/examples/industrial_data_pretraining/paraformer/finetune.shindex 720925223..3041ff399 100644--- a/examples/industrial_data_pretraining/paraformer/finetune.sh+++ b/examples/industrial_data_pretraining/paraformer/finetune.sh@@ -62,5 +62,6 @@ torchrun \++train_conf.validate_interval=2000 \++train_conf.save_checkpoint_interval=2000 \++train_conf.keep_nbest_models=20 \+++train_conf.avg_nbest_model=5 \++optim_conf.lr=0.0002 \++output_dir="${output_dir}" &> ${log_file}\ No newline at end of file