Based on edge-tts encapsulation of speech to text, speech is generated according to text, the same copy is generated only once, real-time generation.
Web UI： https://tts-api.wdft.com
pip3 install edge_tts config.json:
{ "serverURI": "localhost", "localPort": 59443, "audioPath": "/media-files/", "mediaFilesRoot": "/home/wwwroot/tts-service", "validTokenStr": "github.com/ljq" } systemd: tts-service.service
#/etc/systemd/system/tts-service.service sudo systemctl anable tts-service sudo systemctl start tts-service sudo systemctl stop tts-service sudo systemctl restart tts-service sudo systemctl status tts-service Recommended tool for hot update debugging and building: air：
.air.toml
# .air.toml # build [build] cmd = "env GOOS=linux GOARCH=amd64 go build -o ./build/bin/tts-service tts_service.go" # port [run] port = "8081" Audio file deploy web root directory：
mkdir -p /home/wwwroot/tts-service/media-files ### nginx proxy /home/wwwroot/tts-service web request audio path：https://tts-api.wdft.com/media-files/
### GET or POST Method(Nginx proxy) http://localhost:52843/tts?text=txt&role=zh-CN-YunxiNeural&access_token=ljq@GitHub test access_token: 02bd6f28f95dc84cfbf22e0ce1083f77
SERVICE_NAME="tts-service" START_COMMAND="/data/edge-tts/tts-service" nohup /data/edge-tts/tts-service > /dev/null 2>&1 & - yunxi: 【zh-CN-YunxiNeural】
- xiaoyi: 【zh-CN-XiaoyiNeural】
- xiaoxiao: 【zh-CN-XiaoxiaoNeural】
- yunjian: 【zh-CN-YunjianNeural】
- yunxia: 【zh-CN-YunxiaNeural】
- yunyang: 【zh-CN-YunyangNeural】
- LIAONING-xiaobei: 【zh-CN-liaoning-XiaobeiNeural】
- SHNAXI-xiaoni: 【zh-CN-shaanxi-XiaoniNeural】
- HK-xiaojia: 【zh-HK-HiuGaaiNeural】
- HK-chenman: 【zh-HK-HiuMaanNeural】
- HK-yunlong: 【zh-HK-WanLungNeural】
- TW-xiaochen: 【zh-TW-HsiaoChenNeural】
- TW-xiaoyu: 【zh-TW-HsiaoYuNeural】
- TW-yunzhe: 【zh-TW-YunJheNeural】