From adce588c49df248e47e9edfc416df4e52d438269 Mon Sep 17 00:00:00 2001 From: jinzr Date: Wed, 22 Nov 2023 17:41:00 +0800 Subject: [PATCH] minor updates --- egs/multi_zh_en/ASR/zipformer/asr_datamodule.py | 5 +---- egs/multi_zh_en/ASR/zipformer/export.py | 16 ++++++++-------- egs/multi_zh_en/ASR/zipformer/pretrained.py | 16 ++++++++-------- 3 files changed, 17 insertions(+), 20 deletions(-) diff --git a/egs/multi_zh_en/ASR/zipformer/asr_datamodule.py b/egs/multi_zh_en/ASR/zipformer/asr_datamodule.py index 3d58ebf3a..be6e94472 100644 --- a/egs/multi_zh_en/ASR/zipformer/asr_datamodule.py +++ b/egs/multi_zh_en/ASR/zipformer/asr_datamodule.py @@ -19,23 +19,20 @@ import argparse import inspect import logging -from functools import lru_cache from pathlib import Path from typing import Any, Dict, Optional import torch -from lhotse import CutSet, Fbank, FbankConfig, load_manifest, load_manifest_lazy +from lhotse import CutSet, Fbank, FbankConfig, load_manifest from lhotse.dataset import ( # noqa F401 for PrecomputedFeatures CutConcatenate, CutMix, DynamicBucketingSampler, K2SpeechRecognitionDataset, - PrecomputedFeatures, SimpleCutSampler, SpecAugment, ) from lhotse.dataset.input_strategies import ( # noqa F401 For AudioSamples - AudioSamples, OnTheFlyFeatures, ) from lhotse.utils import fix_random_seed diff --git a/egs/multi_zh_en/ASR/zipformer/export.py b/egs/multi_zh_en/ASR/zipformer/export.py index 723288191..fbd9ce0dd 100755 --- a/egs/multi_zh_en/ASR/zipformer/export.py +++ b/egs/multi_zh_en/ASR/zipformer/export.py @@ -33,7 +33,7 @@ dataset, you should change the argument values according to your dataset. ./zipformer/export.py \ --exp-dir ./zipformer/exp \ - --tokens data/lang_bpe_2000/tokens.txt \ + --tokens data/lang_bbpe_2000/tokens.txt \ --epoch 20 \ --avg 1 \ --jit 1 @@ -53,7 +53,7 @@ for how to use the exported models outside of icefall. --causal 1 \ --chunk-size 16 \ --left-context-frames 128 \ - --tokens data/lang_bpe_2000/tokens.txt \ + --tokens data/lang_bbpe_2000/tokens.txt \ --epoch 20 \ --avg 1 \ --jit 1 @@ -72,7 +72,7 @@ for how to use the exported models outside of icefall. ./zipformer/export.py \ --exp-dir ./zipformer/exp \ - --tokens data/lang_bpe_2000/tokens.txt \ + --tokens data/lang_bbpe_2000/tokens.txt \ --epoch 20 \ --avg 1 @@ -81,7 +81,7 @@ for how to use the exported models outside of icefall. ./zipformer/export.py \ --exp-dir ./zipformer/exp \ --causal 1 \ - --tokens data/lang_bpe_2000/tokens.txt \ + --tokens data/lang_bbpe_2000/tokens.txt \ --epoch 20 \ --avg 1 @@ -103,7 +103,7 @@ you can do: --avg 1 \ --max-duration 600 \ --decoding-method greedy_search \ - --bpe-model data/lang_bpe_2000/bpe.model + --bpe-model data/lang_bbpe_2000/bpe.model - For streaming model: @@ -124,7 +124,7 @@ To use the generated file with `zipformer/decode.py` and `zipformer/streaming_de --chunk-size 16 \ --left-context-frames 128 \ --decoding-method greedy_search \ - --bpe-model data/lang_bpe_2000/bpe.model + --bpe-model data/lang_bbpe_2000/bpe.model # chunk-wise streaming decoding ./zipformer/streaming_decode.py \ @@ -136,7 +136,7 @@ To use the generated file with `zipformer/decode.py` and `zipformer/streaming_de --chunk-size 16 \ --left-context-frames 128 \ --decoding-method greedy_search \ - --bpe-model data/lang_bpe_2000/bpe.model + --bpe-model data/lang_bbpe_2000/bpe.model Check ./pretrained.py for its usage. @@ -251,7 +251,7 @@ def get_parser(): parser.add_argument( "--tokens", type=str, - default="data/lang_bpe_2000/tokens.txt", + default="data/lang_bbpe_2000/tokens.txt", help="Path to the tokens.txt", ) diff --git a/egs/multi_zh_en/ASR/zipformer/pretrained.py b/egs/multi_zh_en/ASR/zipformer/pretrained.py index 69ff382da..e2bc6f9ca 100755 --- a/egs/multi_zh_en/ASR/zipformer/pretrained.py +++ b/egs/multi_zh_en/ASR/zipformer/pretrained.py @@ -25,7 +25,7 @@ dataset, you should change the argument values according to your dataset. ./zipformer/export.py \ --exp-dir ./zipformer/exp \ - --tokens data/lang_bpe_2000/tokens.txt \ + --tokens data/lang_bbpe_2000/tokens.txt \ --epoch 23 \ --avg 1 @@ -34,7 +34,7 @@ dataset, you should change the argument values according to your dataset. ./zipformer/export.py \ --exp-dir ./zipformer/exp \ --causal 1 \ - --tokens data/lang_bpe_2000/tokens.txt \ + --tokens data/lang_bbpe_2000/tokens.txt \ --epoch 23 \ --avg 1 @@ -45,7 +45,7 @@ Usage of this script: (1) greedy search ./zipformer/pretrained.py \ --checkpoint ./zipformer/exp/pretrained.pt \ - --tokens data/lang_bpe_2000/tokens.txt \ + --tokens data/lang_bbpe_2000/tokens.txt \ --method greedy_search \ /path/to/foo.wav \ /path/to/bar.wav @@ -53,7 +53,7 @@ Usage of this script: (2) modified beam search ./zipformer/pretrained.py \ --checkpoint ./zipformer/exp/pretrained.pt \ - --tokens ./data/lang_bpe_2000/tokens.txt \ + --tokens ./data/lang_bbpe_2000/tokens.txt \ --method modified_beam_search \ /path/to/foo.wav \ /path/to/bar.wav @@ -61,7 +61,7 @@ Usage of this script: (3) fast beam search ./zipformer/pretrained.py \ --checkpoint ./zipformer/exp/pretrained.pt \ - --tokens ./data/lang_bpe_2000/tokens.txt \ + --tokens ./data/lang_bbpe_2000/tokens.txt \ --method fast_beam_search \ /path/to/foo.wav \ /path/to/bar.wav @@ -74,7 +74,7 @@ Usage of this script: --causal 1 \ --chunk-size 16 \ --left-context-frames 128 \ - --tokens ./data/lang_bpe_2000/tokens.txt \ + --tokens ./data/lang_bbpe_2000/tokens.txt \ --method greedy_search \ /path/to/foo.wav \ /path/to/bar.wav @@ -85,7 +85,7 @@ Usage of this script: --causal 1 \ --chunk-size 16 \ --left-context-frames 128 \ - --tokens ./data/lang_bpe_2000/tokens.txt \ + --tokens ./data/lang_bbpe_2000/tokens.txt \ --method modified_beam_search \ /path/to/foo.wav \ /path/to/bar.wav @@ -96,7 +96,7 @@ Usage of this script: --causal 1 \ --chunk-size 16 \ --left-context-frames 128 \ - --tokens ./data/lang_bpe_2000/tokens.txt \ + --tokens ./data/lang_bbpe_2000/tokens.txt \ --method fast_beam_search \ /path/to/foo.wav \ /path/to/bar.wav