From e66da76db9b24fcfd58e2578052cad5fb896a629 Mon Sep 17 00:00:00 2001 From: huangyuxin Date: Fri, 29 Oct 2021 10:16:09 +0000 Subject: [PATCH] fix the bug of chooing dataloader, remove the log of downloads lm, change the epoch in tiny --- README.md | 18 +++++++++--------- deepspeech/exps/deepspeech2/model.py | 9 ++++++--- examples/aishell/s0/local/test.sh | 2 +- examples/aishell/s0/local/test_export.sh | 2 +- examples/librispeech/s0/local/test.sh | 2 +- examples/other/1xt2x/aishell/local/test.sh | 2 +- examples/other/1xt2x/baidu_en8k/local/test.sh | 2 +- examples/other/1xt2x/librispeech/local/test.sh | 2 +- examples/tiny/s0/conf/deepspeech2.yaml | 2 +- examples/tiny/s0/conf/deepspeech2_online.yaml | 2 +- examples/tiny/s0/local/test.sh | 2 +- examples/tiny/s1/conf/chunk_confermer.yaml | 2 +- examples/tiny/s1/conf/chunk_transformer.yaml | 2 +- examples/tiny/s1/conf/conformer.yaml | 2 +- examples/tiny/s1/conf/transformer.yaml | 2 +- 15 files changed, 28 insertions(+), 25 deletions(-) diff --git a/README.md b/README.md index 7060a655..8a83ac61 100644 --- a/README.md +++ b/README.md @@ -9,20 +9,20 @@ English | [简体中文](README_ch.md)

-

+

Quick Start | Tutorials - | Models List + | Models List

- + ------------------------------------------------------------------------------------ ![License](https://img.shields.io/badge/license-Apache%202-red.svg) ![python version](https://img.shields.io/badge/python-3.7+-orange.svg) ![support os](https://img.shields.io/badge/os-linux-yellow.svg) @@ -31,7 +31,7 @@ how they can use it Via the easy-to-use, efficient, flexible and scalable implementation, our vision is to empower both industrial application and academic research, including training, inference & testing modules, and deployment process. To be more specific, this toolkit features at: - **Fast and Light-weight**: we provide high-speed and ultra-lightweight models that are convenient for industrial deployment. - **Rule-based Chinese frontend**: our frontend contains Text Normalization (TN) and Grapheme-to-Phoneme (G2P, including Polyphone and Tone Sandhi). Moreover, we use self-defined linguistic rules to adapt Chinese context. -- **Varieties of Functions that Vitalize both Industrial and Academia**: +- **Varieties of Functions that Vitalize both Industrial and Academia**: - *Implementation of critical audio tasks*: this toolkit contains audio functions like Speech Translation (ST), Automatic Speech Recognition (ASR), Text-To-Speech Synthesis (TTS), Voice Cloning(VC), Punctuation Restoration, etc. - *Integration of mainstream models and datasets*: the toolkit implements modules that participate in the whole pipeline of the speech tasks, and uses mainstream datasets like LibriSpeech, LJSpeech, AIShell, CSMSC, etc. See also [model lists](#models-list) for more details. - *Cross-domain application*: as an extension of the application of traditional audio tasks, we combine the aforementioned tasks with other fields like NLP. @@ -70,7 +70,7 @@ If you want to set up PaddleSpeech in other environment, please see the [ASR ins ## Quick Start > Note: the current links to `English ASR` and `English TTS` are not valid. -Just a quick test of our functions: [English ASR](link/hubdetail?name=deepspeech2_aishell&en_category=AutomaticSpeechRecognition) and [English TTS](link/hubdetail?name=fastspeech2_baker&en_category=TextToSpeech) by typing message or upload your own audio file. +Just a quick test of our functions: [English ASR](link/hubdetail?name=deepspeech2_aishell&en_category=AutomaticSpeechRecognition) and [English TTS](link/hubdetail?name=fastspeech2_baker&en_category=TextToSpeech) by typing message or upload your own audio file. Developers can have a try of our model with only a few lines of code. @@ -87,7 +87,7 @@ bash local/test.sh conf/deepspeech2.yaml ckptfile offline ``` For *TTS*, try FastSpeech2 on LJSpeech: -- Download LJSpeech-1.1 from the [ljspeech official website](https://keithito.com/LJ-Speech-Dataset/), our prepared durations for fastspeech2 [ljspeech_alignment](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz). +- Download LJSpeech-1.1 from the [ljspeech official website](https://keithito.com/LJ-Speech-Dataset/), our prepared durations for fastspeech2 [ljspeech_alignment](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz). - The pretrained models are seperated into two parts: [fastspeech2_nosil_ljspeech_ckpt](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_ljspeech_ckpt_0.5.zip) and [pwg_ljspeech_ckpt](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_ljspeech_ckpt_0.5.zip). Please download then unzip to `./model/fastspeech2` and `./model/pwg` respectively. - Assume your path to the dataset is `~/datasets/LJSpeech-1.1` and `./ljspeech_alignment` accordingly, preprocess your data and then use our pretrained model to synthesize: ```shell @@ -106,7 +106,7 @@ PaddleSpeech supports a series of most popular models, summarized in [released m ASR module contains *Acoustic Model* and *Language Model*, with the following details: > Note: The `Link` should be code path rather than download links. diff --git a/deepspeech/exps/deepspeech2/model.py b/deepspeech/exps/deepspeech2/model.py index 6424cfdf..710630a7 100644 --- a/deepspeech/exps/deepspeech2/model.py +++ b/deepspeech/exps/deepspeech2/model.py @@ -153,8 +153,12 @@ class DeepSpeech2Trainer(Trainer): def setup_model(self): config = self.config.clone() with UpdateConfig(config): - config.model.feat_size = self.train_loader.collate_fn.feature_size - config.model.dict_size = self.train_loader.collate_fn.vocab_size + if self.train: + config.model.feat_size = self.train_loader.collate_fn.feature_size + config.model.dict_size = self.train_loader.collate_fn.vocab_size + else: + config.model.feat_size = self.test_loader.collate_fn.feature_size + config.model.dict_size = self.test_loader.collate_fn.vocab_size if self.args.model_type == 'offline': model = DeepSpeech2Model.from_config(config.model) @@ -189,7 +193,6 @@ class DeepSpeech2Trainer(Trainer): self.lr_scheduler = lr_scheduler logger.info("Setup optimizer/lr_scheduler!") - def setup_dataloader(self): config = self.config.clone() config.defrost() diff --git a/examples/aishell/s0/local/test.sh b/examples/aishell/s0/local/test.sh index 2ae0740b..64d72503 100755 --- a/examples/aishell/s0/local/test.sh +++ b/examples/aishell/s0/local/test.sh @@ -13,7 +13,7 @@ ckpt_prefix=$2 model_type=$3 # download language model -bash local/download_lm_ch.sh +bash local/download_lm_ch.sh > dev/null 2>&1 if [ $? -ne 0 ]; then exit 1 fi diff --git a/examples/aishell/s0/local/test_export.sh b/examples/aishell/s0/local/test_export.sh index a9a6b122..71469753 100755 --- a/examples/aishell/s0/local/test_export.sh +++ b/examples/aishell/s0/local/test_export.sh @@ -13,7 +13,7 @@ jit_model_export_path=$2 model_type=$3 # download language model -bash local/download_lm_ch.sh +bash local/download_lm_ch.sh > dev/null 2>&1 if [ $? -ne 0 ]; then exit 1 fi diff --git a/examples/librispeech/s0/local/test.sh b/examples/librispeech/s0/local/test.sh index 4d00f30b..25dd0437 100755 --- a/examples/librispeech/s0/local/test.sh +++ b/examples/librispeech/s0/local/test.sh @@ -13,7 +13,7 @@ ckpt_prefix=$2 model_type=$3 # download language model -bash local/download_lm_en.sh +bash local/download_lm_en.sh > /dev/null 2>&1 if [ $? -ne 0 ]; then exit 1 fi diff --git a/examples/other/1xt2x/aishell/local/test.sh b/examples/other/1xt2x/aishell/local/test.sh index 2ae0740b..d539ac49 100755 --- a/examples/other/1xt2x/aishell/local/test.sh +++ b/examples/other/1xt2x/aishell/local/test.sh @@ -13,7 +13,7 @@ ckpt_prefix=$2 model_type=$3 # download language model -bash local/download_lm_ch.sh +bash local/download_lm_ch.sh > /dev/null 2>&1 if [ $? -ne 0 ]; then exit 1 fi diff --git a/examples/other/1xt2x/baidu_en8k/local/test.sh b/examples/other/1xt2x/baidu_en8k/local/test.sh index 4d00f30b..25dd0437 100755 --- a/examples/other/1xt2x/baidu_en8k/local/test.sh +++ b/examples/other/1xt2x/baidu_en8k/local/test.sh @@ -13,7 +13,7 @@ ckpt_prefix=$2 model_type=$3 # download language model -bash local/download_lm_en.sh +bash local/download_lm_en.sh > /dev/null 2>&1 if [ $? -ne 0 ]; then exit 1 fi diff --git a/examples/other/1xt2x/librispeech/local/test.sh b/examples/other/1xt2x/librispeech/local/test.sh index 4d00f30b..25dd0437 100755 --- a/examples/other/1xt2x/librispeech/local/test.sh +++ b/examples/other/1xt2x/librispeech/local/test.sh @@ -13,7 +13,7 @@ ckpt_prefix=$2 model_type=$3 # download language model -bash local/download_lm_en.sh +bash local/download_lm_en.sh > /dev/null 2>&1 if [ $? -ne 0 ]; then exit 1 fi diff --git a/examples/tiny/s0/conf/deepspeech2.yaml b/examples/tiny/s0/conf/deepspeech2.yaml index 621b372c..58899a15 100644 --- a/examples/tiny/s0/conf/deepspeech2.yaml +++ b/examples/tiny/s0/conf/deepspeech2.yaml @@ -45,7 +45,7 @@ model: ctc_grad_norm_type: null training: - n_epoch: 10 + n_epoch: 5 accum_grad: 1 lr: 1e-5 lr_decay: 0.8 diff --git a/examples/tiny/s0/conf/deepspeech2_online.yaml b/examples/tiny/s0/conf/deepspeech2_online.yaml index 5a8294ad..334b1d31 100644 --- a/examples/tiny/s0/conf/deepspeech2_online.yaml +++ b/examples/tiny/s0/conf/deepspeech2_online.yaml @@ -47,7 +47,7 @@ model: ctc_grad_norm_type: null training: - n_epoch: 10 + n_epoch: 5 accum_grad: 1 lr: 1e-5 lr_decay: 1.0 diff --git a/examples/tiny/s0/local/test.sh b/examples/tiny/s0/local/test.sh index 4d00f30b..25dd0437 100755 --- a/examples/tiny/s0/local/test.sh +++ b/examples/tiny/s0/local/test.sh @@ -13,7 +13,7 @@ ckpt_prefix=$2 model_type=$3 # download language model -bash local/download_lm_en.sh +bash local/download_lm_en.sh > /dev/null 2>&1 if [ $? -ne 0 ]; then exit 1 fi diff --git a/examples/tiny/s1/conf/chunk_confermer.yaml b/examples/tiny/s1/conf/chunk_confermer.yaml index b14b4b21..c5186669 100644 --- a/examples/tiny/s1/conf/chunk_confermer.yaml +++ b/examples/tiny/s1/conf/chunk_confermer.yaml @@ -83,7 +83,7 @@ model: training: - n_epoch: 20 + n_epoch: 5 accum_grad: 1 global_grad_clip: 5.0 optim: adam diff --git a/examples/tiny/s1/conf/chunk_transformer.yaml b/examples/tiny/s1/conf/chunk_transformer.yaml index 38edbf35..29c30b26 100644 --- a/examples/tiny/s1/conf/chunk_transformer.yaml +++ b/examples/tiny/s1/conf/chunk_transformer.yaml @@ -76,7 +76,7 @@ model: training: - n_epoch: 20 + n_epoch: 5 accum_grad: 1 global_grad_clip: 5.0 optim: adam diff --git a/examples/tiny/s1/conf/conformer.yaml b/examples/tiny/s1/conf/conformer.yaml index 0b06b2b7..8487da77 100644 --- a/examples/tiny/s1/conf/conformer.yaml +++ b/examples/tiny/s1/conf/conformer.yaml @@ -79,7 +79,7 @@ model: training: - n_epoch: 20 + n_epoch: 5 accum_grad: 4 global_grad_clip: 5.0 optim: adam diff --git a/examples/tiny/s1/conf/transformer.yaml b/examples/tiny/s1/conf/transformer.yaml index 1c6f9e02..cc9b5c51 100644 --- a/examples/tiny/s1/conf/transformer.yaml +++ b/examples/tiny/s1/conf/transformer.yaml @@ -73,7 +73,7 @@ model: training: - n_epoch: 21 + n_epoch: 5 accum_grad: 1 global_grad_clip: 5.0 optim: adam -- GitLab