MIT-licensed DSH plugin for agent-led audio: sampled instruments, timeline arranging, sound effects, TTS, automatic mix balance and a visual DAW — capped at 15 minutes and 20 tracks per project.
DSH integration
Native runtime
Author-claimed
Safety audit
Unaudited
Last verified
2026-11-11
License
MIT
01What can it help you accomplish?
Turn a sound idea into a finished audio project from a DSH conversation
A rendered multitrack project — sampled instruments, speech, sound effects — arranged on one timeline and mixed to a finished file
Video creators, podcasters and game-audio authors who direct an agent in DSH instead of hand-writing synthesis code
Audition and fine-tune what the agent produced, track by track
A visual DAW with per-track waveforms, draggable clips, clip gain and fades, plus "Add to conversation" so a selected range becomes a reference for the agent
People who want to fix a late line or a buried effect by hand instead of regenerating the whole mix
02How to install into DeepSeek Harness
Prerequisites
- Node.js 22.19+ (22.x) or 24+, which ships with DSH
- ffmpeg — only needed for speech and for muxing video
- A SoundFont bank installed from Settings; the repository bundles none
Installation steps
- 01
Run `dsh plugin --profile desktop add dsh-any-audio` (replace `desktop` with the profile DSH actually boots)
$ dsh plugin --profile desktop add dsh-any-audio
- 02
Restart DSH once so it picks up the plugin
- 03
Open the AnyAudio page in DSH Settings to install a sound source (GeneralUser GS is the default suggestion), switch the speech engine and adjust mixing defaults
Verify the integration
Not specified by the author
03DSH integration and capability boundaries
DSH plugin bundle installed through the dsh CLI, with an AnyAudio settings page and a DAW tab inside DSH
Sampled instrument rendering
notes written by the agent→instrument audio rendered from local SoundFont banks (.sf2 / .sf3 / .dls) in `~/.dsh/anyaudio/fonts`
one shared sound library is used by every project and agent session on the machineTimeline arranging
rendered clips, speech lines, effects→one timeline with millisecond placement shared by music, narration and effects
a project is capped at 15 minutes and 20 tracksAutomatic mix balance
track roles (e.g. `narration`), gain, pan and fades→ordered tracks with per-track gain/pan/fades, ducking under narration, and a peak-normalised master
declaring a track as narration automatically ducks the other busesSpeech synthesis
dialogue lines→speech via Edge TTS (default, no key) or MiMo-V2.5-TTS (opt-in, Open Platform API key)
Edge TTS goes through Microsoft's public endpoint, needs ffmpeg to convert MP3 to WAV, and may be rate-limitedthe MiMo key is stored in DSH credentials and only read on the host sideHeadless renderer
a score.json file→a wav file, without an agent in the loop
04Who is it for? When not to use it?
Good for
- Video creators, podcasters and game-audio authors who direct an agent in DSH instead of hand-writing synthesis code
- People who want to fix a late line or a buried effect by hand instead of regenerating the whole mix
Not for
- A project is at most 15 minutes long and 20 tracks wide; exceeding either is an error.
- Video can sit in the project as a reference picture, but this version does not trim, join or transition it.
- SoundFont only — no SFZ, Kontakt or VST, and converting other source formats is out of scope for this version.
05Compatibility, maintenance and safety notes
- A project is at most 15 minutes long and 20 tracks wide; exceeding either is an error.
- Video can sit in the project as a reference picture, but this version does not trim, join or transition it.
- SoundFont only — no SFZ, Kontakt or VST, and converting other source formats is out of scope for this version.
- No SoundFont is bundled with the package: you install a bank from Settings or drop compatible files into `~/.dsh/anyaudio/fonts` first.
- There is no per-note envelope control: attack and release come from the preset, so staccato relies on short notes with gaps and dynamics on velocity plus expression.
- Chinese instruments are General MIDI approximations — `guzheng` maps to Koto, `dizi` to Flute, `erhu` to Fiddle — so dedicated banks will sound better.
- Tracks hold finished audio only: every clip is an already-rendered wav/mp3, and MIDI, scores and parameters never go on the timeline.
MIT · README evidence only, latest release v0.2.0, safety not audited
06Frequently asked questions
How do I install AnyAudio in DeepSeek Harness?
Run `dsh plugin --profile desktop add dsh-any-audio`, replacing `desktop` with the profile DSH actually boots, then restart DSH once. After that, open the AnyAudio page in DSH Settings to install a sound source, pick the speech engine and set the mixing defaults.
Do I need to bring my own SoundFont?
Yes — the repository bundles no SoundFonts. The plugin suggests GeneralUser GS as the default foundation and MuseScore General as an extra orchestral bank, and it can also load compatible `.sf2`, `.sf3` or `.dls` files placed in `~/.dsh/anyaudio/fonts`, which is one shared library for every project on the machine.
Is any API key required?
Not for the default path: Edge TTS goes through Microsoft's public endpoint and needs ffmpeg to convert MP3 to WAV, and may be rate-limited. MiMo-V2.5-TTS is opt-in with a normal Open Platform API key stored in DSH credentials and read only on the host side.
What are the project limits?
A project is at most 15 minutes long and 20 tracks wide, and going past either is an error. Tracks hold finished wav/mp3 audio — MIDI, scores and parameters never go on the timeline — and video can only sit in the project as a reference picture.
How do I fix a line that lands half a second late?
Open the DAW tab and move the clip by hand — the README recommends that over regenerating, because person and agent share one timeline. Select a range and press "Add to conversation" to make it a reference, then describe the change you want.
07Related DSH workflows
dsh-our-free-model
by ebony-vinyl
Adds model choices, availability checks, usage panels, and a local OpenAI-compatible forwarder.
dsh-routing-suite
by yjh051108
dsh-routing-suite — injector + router-standard kit: install the runtime injector first, then the task-aware reasoning-mode router preset (measured P1-P23).
dsh-ops
by t-auto
Run Bash and PowerShell commands, search and edit local files, and manage long-running background jobs from DSH.
brooks-lint
by hyhmrright
Portable AI code quality reviews for Codex, Cursor, Copilot and other coding agents: six Agent Skills for architecture, technical debt, tests and remediation.
08Data and sources
AnyAudio is an audio production toolkit for AI agents.
dsh plugin --profile desktop add dsh-any-audio
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-11-11. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.