如何使用ModelScope魔搭开源代码训练一款语音合成模型

git clone https://github.com/AlibabaResearch/KAN-TTS.gitcd KAN-TTS

# 防止使用pip安装时出现网络问题，建议切换国内pip源pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
# 创建虚拟环境conda env create -f environment.yaml
# 激活虚拟环境conda activate maas

.├── interval│   ├── 500001.interval│   ├── 500002.interval│   ├── 500003.interval│   ├── ...│   └── 600010.interval├── prosody│   └── prosody.txt└── wav    ├── 500001.wav    ├── 500002.wav    ├── ...    └── 600010.wav

.├── prosody│   └── prosody.txt└── wav    ├── 1.wav    ├── 2.wav    ├── ...    └── 9000.wav

unzip aishell3.zip

# 拉取元数据仓库，并做重采样处理git clone https://www.modelscope.cn/datasets/speech_tts/AISHELL-3.git./AISHELL-3/aishell_resample.sh aishell3 aishell3_16k 16000

python kantts/preprocess/data_process.py --voice_input_dir aishell3_16k/SSB0009 --voice_output_dir training_stage/SSB0009_feats --audio_config kantts/configs/audio_config_16k.yaml --speaker SSB0009

# 基于阿里标准格式数据所提取出的特征目录├── am_train.lst├── am_valid.lst├── audio_config.yaml├── badlist.txt├── data_process_stdout.log├── duration├── energy├── f0├── frame_energy├── frame_f0├── frame_uv├── mel├── raw_duration├── raw_metafile.txt├── Script.xml├── train.lst├── valid.lst└── wav

# 克隆预训练模型git clone -b pretrain http://www.modelscope.cn/speech_tts/speech_sambert-hifigan_tts_zh-cn_multisp_pretrain_16k.git

# 将speaker_list其中一个发音人替换为aishell3发音人linguistic_unit:   cleaners: english_cleaners  lfeat_type_list: sy,tone,syllable_flag,word_segment,emo_category,speaker_category  speaker_list: SSB0009,F74,FBYN,FRXL,M7,xiaoyu

# 将train_max_steps改为1100100...train_max_steps: 1100100...

CUDA_VISIBLE_DEVICES=0 python kantts/bin/train_sambert.py --model_config speech_sambert-hifigan_tts_zh-cn_multisp_pretrain_16k/basemodel_16k/sambert/config.yaml  --root_dir training_stage/SSB0009_feats --stage_dir training_stage/SSB0009_sambert_ckpt

./SSB0009_sambert_ckpt├── ckpt/├── checkpoint_1100000.pth   <----- this is the latest checkpoint├── checkpoint_1020000.pth├── checkpoint_1040000.pth|....├── config.yaml├── log/└── stdout.log

# 将train_max_steps改为100100...train_max_steps: 100100...

CUDA_VISIBLE_DEVICES=0 python kantts/bin/train_hifigan.py --model_config speech_sambert-hifigan_tts_zh-cn_multisp_pretrain_16k/basemodel_16k/hifigan/config.yaml --root_dir training_stage/SSB0009_feats --stage_dir training_stage/SSB0009_hifigan_ckpt

.├── ckpt/│   ├── checkpoint_2020000.pth│   ├── checkpoint_2040000.pth│   ├── ...│   └── checkpoint_2100000.pth      <---- this is the latest checkpoint├── config.yaml├── log/└── stdout.log

徐玠诡谲多智，善揣摩，知道徐知询不可辅佐，掌握着他的短处以归附徐知诰。许乐夫生于山东省临朐县杨善镇大辛庄，毕业于抗大一分校。宣统元年（1909年），顺德绅士冯国材在香山大黄圃成立安洲农务分会，管辖东海十六沙，冯国材任总理。学生们大多住在校区宿舍，通过参加不同的体育文化俱乐部及社交活动，形成一个友谊长存的社会圈。学校的“三节一会”（艺术节、社团节、科技节、运动会）是显示青春才华的盛大活动。雪是先天自闭症患者，不懂与人沟通，却拥有灵敏听觉，而且对复杂动作过目不忘。勋章通过一柱状螺孔和螺钉附着在衣物上。

CUDA_VISIBLE_DEVICES=0 python kantts/bin/text_to_wav.py --txt test.txt --output_dir res/SSB0009_syn --res_zip speech_sambert-hifigan_tts_zh-cn_multisp_pretrain_16k/resource.zip --am_ckpt training_stage/SSB0009_sambert_ckpt/ckpt/checkpoint_1100000.pth --voc_ckpt training_stage/SSB0009_hifigan_ckpt/ckpt/checkpoint_2100000.pth --speaker SSB0009

│   ├── 8_0_dur.txt│   ├── 8_0_energy.txt│   ├── 8_0_f0.txt│   ├── 8_0_mel.npy│   ├── 8_1_dur.txt│   ├── 8_1_energy.txt│   ├── 8_1_f0.txt│   └── 8_1_mel.npy├── res_wavs│   ├── 0.wav│   ├── 1.wav│   ├── 2.wav│   ├── 3.wav│   ├── 4.wav│   ├── 5.wav│   ├── 6.wav├── stdout.log└── symbols.lst