ESPnet

---

关于 ESPnet

• github： https://github.com/espnet/espnet

ESPnet is an end-to-end speech processing toolkit covering end-to-end speech recognition, text-to-speech, speech translation, speech enhancement, speaker diarization, spoken language understanding, and so on.
ESPnet uses pytorch as a deep learning engine and also follows Kaldi style data processing, feature extraction/format, and recipes to provide a complete setup for various speech processing experiments.

ESPnet 是一个端到端语音处理工具包，包含语音识别、文字转语音、语音翻译、语音增强、说话人识别、口语理解等。
ESPnet 使用 PyTorch 作为深度学习引擎，并遵循 Kaldi 风格数据处理、特征抽取/格式、方案，为各种语音处理实验提供完整的设置。

---

安装配置

1、下载

git clone https://github.com/espnet/espnet.git

---

2、设置软链接

cd espnet/tools
ln -s <path to kaldi> .

---

3、安装依赖包

pip install chainer==6.0.0 cupy-cuda92==6.0.0

---

espnet/tools下执行check_install.py

python3 check_install.py

---

4、make

make KALDI=~/xxcode/kaldi PYTHON=~/miniconda3/bin/python CUDA_VERSION=11.3

---

运行 yesno

进入 espnet/egs/yesno 文件夹，下面有 tts1 和 asr1 文件夹。进入一个，然后执行：

sh run.sh

---

tts 执行成功后，打印如下：

Succeeded creating wav for test_yesno
Succeeded creating wav for train_dev
Finished.

---

asr 执行成功后，将打印如下：

2023-01-28 19:57:40,756 (json2trn:46) INFO: reading exp/train_nodev_pytorch_train/decode_test_yesno_decode/data.json
2023-01-28 19:57:40,756 (json2trn:50) INFO: reading data/lang_1char/train_nodev_units.txt
write a CER (or TER) result in exp/train_nodev_pytorch_train/decode_test_yesno_decode/result.txt
       | SPKR   | # Snt # Wrd | Corr    Sub    Del    Ins    Err  S.Err |
       | Sum/Avg|   30    835 | 47.9   51.1    1.0   47.2   99.3  100.0 |
Finished

---

如果执行失败，如果是某个文件、command 找不到，可以手动查找下。如果有就将其所在文件夹添加到环境变量。如果没有，需要检查下，是否某个步骤没有编译成功。
如果这些都没问题， 可以检查下，Kaldi 是否安装配置成功。
Kaldi 安装配置可参考：https://blog.csdn.net/lovechris00/article/details/128347128

---

2023-01-28(周六)
初七、开工第一天，伊织祝大家学有所成