From d7e2259ccf4c3b7bbcd1afc6b73f40e0cc924151 Mon Sep 17 00:00:00 2001 From: "haoneng.lhn" Date: Mon, 25 Sep 2023 16:47:09 +0800 Subject: [PATCH] update paraformer online recipe --- .../demo.py | 2 +- .../demo_online.py | 2 +- .../demo_online_v2.py | 7 +++++-- .../finetune.py | 2 +- .../infer.py | 2 +- 5 files changed, 9 insertions(+), 6 deletions(-) diff --git a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo.py b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo.py index bef38492c..a3d522211 100644 --- a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo.py +++ b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo.py @@ -4,7 +4,7 @@ from modelscope.utils.constant import Tasks inference_pipeline = pipeline( task=Tasks.auto_speech_recognition, model='damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online', - model_revision='v1.0.6', + model_revision='v1.0.7', update_model=False, mode="paraformer_fake_streaming" ) diff --git a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo_online.py b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo_online.py index c1f4afe01..d9b5b2dfb 100644 --- a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo_online.py +++ b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo_online.py @@ -14,7 +14,7 @@ os.environ["MODELSCOPE_CACHE"] = "./" inference_pipeline = pipeline( task=Tasks.auto_speech_recognition, model='damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online', - model_revision='v1.0.6', + model_revision='v1.0.7', update_model=False, mode="paraformer_streaming" ) diff --git a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo_online_v2.py b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo_online_v2.py index 65881cf1e..1f2500d48 100644 --- a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo_online_v2.py +++ b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/demo_online_v2.py @@ -24,9 +24,12 @@ speech, sample_rate = soundfile.read(os.path.join(model_dir, "example/asr_exampl speech_length = speech.shape[0] sample_offset = 0 -chunk_size = [0, 8, 4] #[5, 10, 5] 600ms, [8, 8, 4] 480ms +chunk_size = [0, 10, 5] #[0, 10, 5] 600ms, [0, 8, 4] 480ms +encoder_chunk_look_back = 4 #number of chunks to lookback for encoder self-attention +decoder_chunk_look_back = 1 #number of encoder chunks to lookback for decoder cross-attention stride_size = chunk_size[1] * 960 -param_dict = {"cache": dict(), "is_final": False, "chunk_size": chunk_size, "encoder_chunk_look_back": 4, "decoder_chunk_look_back": 1} +param_dict = {"cache": dict(), "is_final": False, "chunk_size": chunk_size, + "encoder_chunk_look_back": encoder_chunk_look_back, "decoder_chunk_look_back": decoder_chunk_look_back} final_result = "" for sample_offset in range(0, speech_length, min(stride_size, speech_length - sample_offset)): diff --git a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/finetune.py b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/finetune.py index 30034aa25..fa046d8b0 100644 --- a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/finetune.py +++ b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/finetune.py @@ -14,7 +14,7 @@ def modelscope_finetune(params): ds_dict = MsDataset.load(params.data_path) kwargs = dict( model=params.model, - model_revision='v1.0.6', + model_revision='v1.0.7', update_model=False, data_dir=ds_dict, dataset_type=params.dataset_type, diff --git a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/infer.py b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/infer.py index 241ebef22..0537ddafe 100644 --- a/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/infer.py +++ b/egs_modelscope/asr/paraformer/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-online/infer.py @@ -11,7 +11,7 @@ def modelscope_infer(args): model=args.model, output_dir=args.output_dir, batch_size=args.batch_size, - model_revision='v1.0.6', + model_revision='v1.0.7', update_model=False, mode="paraformer_fake_streaming", param_dict={"decoding_model": args.decoding_mode, "hotword": args.hotword_txt}