dsh-plugin-speech

English | 中文

Read assistant replies aloud in DeepSeek Harness, with the audio streamed while it is still being generated.

What it does

Each finished assistant message gets a play control in its action row. Press it and the reply is spoken through a text-to-speech service; press it again and it stops. A settings row in General owns the service, the voice, and the tuning.

The reply is sent in one request and the audio is played as it arrives, so speech starts on the first frames instead of waiting for the whole file. The equivalent of watching an answer stream in, for audio.

Providers

Adding a provider means adding one folder under src/providers/ and one line in src/providers/registry.ts — the settings screen builds its fields from what the provider declares.

Requirements

• A DeepSeek Harness Web surface (dsh web), version 0.1.5-rc.2 or newer.
• Network access to the speech service. Nothing else: there is no model to download, and the Edge provider needs no account.

Install

1. Add the package to your harness checkout:

   pnpm add -w @nakamuraia/dsh-plugin-speech

   Installing from GitHub works the same way, if you would rather track the source:

   pnpm add -w github:NakamuraIA/dsh-plugin-speech

2. Copy speech.patch.yml next to the checkout and start the Web surface with it:

   dsh web --patch ./speech.patch.yml

The patch adds one Loader row. The row name is the package name and also the browser module id the built bundle registers under, so the two stay in step — do not rename one without the other.

Settings

Settings opens on General, where a Read aloud row owns:

• Service — which provider speaks.
• Language and Voice — the catalog entry to use.
• Speed, Volume, Pitch — sent to the service with every request.
• Skip code — leave fenced code and inline code out of the reading.
• Test voice — speaks a sample through the settings on screen, saved or not.

Code blocks, tables, and pasted spreadsheets are never read; links keep their label and lose their target; emoji, quotes, and standalone symbols are dropped. Sentence punctuation is kept, because the voice uses it for pauses.

How it works

browser                    host                        service
───────                    ────                        ───────
click ─ ▶ POST /speech/speak { text }
                           select provider from
                           the durable settings
                           ────────────────────────── ▶ synthesize
        ◀ ─ audio bytes stream back ─ ─ ─ ─ ─ ─ ─ ─ ─ ┘
play as they arrive

Synthesis lives on the host because a browser cannot hold an API key and cannot reach most services across CORS. The browser only posts text and plays bytes.

Development

The source is developed inside a DeepSeek Harness checkout, where the type packages and the client build preset live; this repository carries the source and the built artifacts. To rebuild, copy src/ into packages/client/ui-speech/ of a checkout, rename the package there, and run its build.

License

MIT. The Edge provider drives Microsoft's read-aloud endpoint through msedge-tts (MIT); that endpoint is not a documented, supported Microsoft API, so treat availability as best effort.