A command for speech and audio work through Modellix, including text-to-speech, speech-to-text, and speech-to-speech. Speech-to-text turns recorded speech into written text; text-to-speech creates spoken audio from text.

Use it to create speech, transcribe a public audio file, or clone a voice when the required model, audio, and text details are provided.

It routes different audio tasks to suitable models and checks the required inputs before running them. It also waits for the result and downloads the output.

audio — Command by Modellix

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add commands/modellix/modellix-plugin/audio

Clone the repo

git clone --depth 1 https://github.com/Modellix/modellix-plugin

Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 402 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00015	$0.00402
Opus 5	$0.00008	$0.00201
Sonnet 5	$0.00003	$0.00080
Haiku 4.5	$0.00002	$0.00040

Measured yesterday against content hash 97fe9bb8a794, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

audio scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

commands/audio.md · 30 lines

What it actually says

description: Synthesize, transcribe, or transform speech and audio with Modellix. argument-hint: [tts|stt|sts] [text or public audio URL] disable-model-invocation: true

Run an audio workflow with Modellix. Request: $ARGUMENTS

Follow the Modellix skill (skills/modellix/SKILL.md) for execution policy, credentials, schema lookup, and error handling. This command only fixes the routing.

If $ARGUMENTS is empty or the intended audio workflow is ambiguous, ask whether the user wants text-to-speech, speech-to-text, or speech-to-speech. Never invent text, a voice, or an input URL.
Pick the model:
- A slug named by the user wins; confirm its schema with modellix-cli model describe <slug> --json.
- Text-to-speech → alibaba/qwen-audio-3.0-tts-flash; requires text and a Flash-compatible voice verified against the live model doc.
- Speech-to-text → openai/whisper-1; requires one public audio url.
- Speech-to-speech / voice clone → alibaba/cosyvoice-clone; requires the target CosyVoice model, reference audio url, and synthesis text.
Submit, wait, and persist:

modellix-cli model run \
  --model-slug <slug> \
  --body '<json>' \
  --wait --timeout 10m --json

modellix-cli task download <task_id> --output-dir ./outputs --json

One paid submit per invocation. If the outcome is unknown or ambiguous, run modellix-cli task history and recover the existing task—never repeat the same submission blindly.
Report the model slug, task id, and either the local audio paths or the returned transcript resource.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

yesterday First seen · 30 lines · 0 tokens per session scan A 97fe9bb8a794

Subscribe to this mod's changes

audio is a command published in the GitHub repository Modellix/modellix-plugin (1 stars, last pushed 21d ago), licensed MIT. It adds 15 tokens to every session and 402 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Creative and media Data and AI agent-plugin agent-skills claude-code-plugin claude-skills clawhub-skills codex-plugin

audio

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

What it actually says

description: Synthesize, transcribe, or transform speech and audio with Modellix. argument-hint: [tts|stt|sts] [text or public audio URL] disable-model-invocation: true

What this file has done since we first saw it

Other commands, from other repositories

codex-swarm

codex-imagegen

stt

audition-voices

status

music-suno-prompt