dsh-tts：用 OpenAI 兼容后端把文本合成语音
dsh-tts 是 xuan25 维护的 DSH out-of-tree 插件，提供单一 tts 工具，经 OpenAI 兼容后端将文本合成为 wav/mp3 等音频，并支持 voices、providers、models、status、phonemize 等 verb。配置可通过 TTS_* 环境变量或 settings.yaml 的 dsh-tts 段分层覆盖。安装命令为 dsh plugin --profile <name> add dsh-tts。适合需要在智能体流水线中统一对接多种 TTS 后端的开发者；使用前请查阅源码与 MIT 许可证，并注意插件以 dsh 进程权限运行。
Read Moredsh-tts: Turn Text into Speech, Keep the Engine Outside the Plugin
The `dsh-tts` plugin decouples the TTS engine from synthesis logic, focusing on audio file generation and timestamp calculation. It supports two providers: Microsoft Edge (default, with word-level timestamps) and a local command-line interface (offline, configuration required). Key features include synthesizing single utterances and multi-speaker conversations, environment checks, and documentation lookup. It outputs audio files and a `.words.json` sidecar file for downstream use. Installation is convenient, but commercial licensing for Edge and limitations on Chinese word-level timestamps and generation speed in local engines should be noted.
Read MoreDeepSeek Harness TTS Plugin: Multi-Provider Text-to-Speech with a Fallback Chain
The `dsh-tts` plugin for DeepSeek Harness provides real-time text-to-speech (TTS) capabilities for agent responses through the Web UI. It supports a multi-provider fallback chain spanning from offline engines (such as Piper and Edge TTS) to cloud APIs, featuring low-latency streaming audio delivery and a pronunciation dictionary for IT terminology. The plugin also supports Messenger integration, audio export, and multi-agent persona configuration. Installation requires specific dependencies, while neural model inference currently needs separate configuration. API keys are handled solely on the backend, ensuring privacy and security without compromising performance.
Read Moredsh-tts: Voice broadcast plugin for dsh web final results
dsh-tts is a dsh web text-to-speech (TTS) plugin developed by vim0x3c (version 0.1.0, MIT license), designed to reduce the reading cost of long text through voice synthesis. Utilizing Edge TTS technology, this plugin reads only the final assistant response while ignoring intermediate steps. It supports segmented serial reading of long replies, speed adjustment ranging from 0.5x to 2x, and offers over 20 voice options. It provides built-in approval sounds or custom audio functionality and automatically falls back to the browser's local Web Speech API when Edge cloud synthesis fails. The plugin has no third-party dependencies, supports offline installation, and requires installation into the web profile via a local tarball combined with the `-w` flag. After installation, the local bridge service binds to 127.0.0.1 and occupies ports 8765-8780 to provide synthesis and health check interfaces, setting...
Read More