voice-clone

A guided workflow for creating and saving a cloned synthetic voice for a brand. It uses reference audio to test a Qwen3-TTS voice and stores the resulting profile for later use.

In plain words
What is it for?
It helps verify required RunPod settings, record or import reference audio, choose a brand, test the cloned voice, and save a profile that can be loaded automatically.
Why use it?
It brings the setup checks, brand selection, testing, and saving steps into one process. This reduces the chance of missing required service settings or overwriting an existing voice profile without checking it.

Command for Claude Code

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/digitalsamba/claude-code-video-toolkit/voice-clone
Clone the repo
git clone --depth 1 https://github.com/digitalsamba/claude-code-video-toolkit

Made for: Claude Code.

Per session 13 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,868 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.

ModelPer sessionOnce invoked
Fable 5 $0.00013 $0.01868
Opus 5 $0.00006 $0.00934
Sonnet 5 $0.00003 $0.00374
Haiku 4.5 $0.00001 $0.00187

Measured 3d ago against content hash 02c853057892, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

voice-clone 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 3d ago.

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.

.claude/commands/voice-clone.md · 215 lines

How it starts

The opening of the file, as written. The whole thing — 215 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Voice Clone

Guided workflow to set up Qwen3-TTS voice cloning for a brand. Records or imports reference audio, tests the clone, and saves the profile so --brand loads it automatically.

Step 1: Check Dependencies

Verify the environment is ready:

1. Check .env for RUNPOD_API_KEY
   - If missing: "Add `RUNPOD_API_KEY=your_key` to `.env`"
2. Check .env for RUNPOD_QWEN3_TTS_ENDPOINT_ID
   - If missing and API key exists: offer to run `uv run tools/qwen3_tts.py --setup`
   - If API key also missing: guide user to add it first
3. Only proceed once both are confirmed

Use Grep to check .env for both keys. If the endpoint is missing but the API key exists, offer to run setup automatically.

Step 2: Choose Brand

Scan for available brands:

1. Glob brands/*/brand.json
2. For each brand, check if voice.json exists and has a qwen3.clone section
3. Present list:

   Available brands:

     1. default — no clone profile
     2. digital-samba — clone configured (assets/voice-reference.m4a)

   Which brand should this voice clone be saved to?

If the selected brand already has a clone profile, show the existing config and ask:

  • Test existing clone — generate a sample to hear it
  • Replace — record/import new reference audio
  • Cancel — exit

Step 3: Record or Provide Reference Audio

Present a reference script for the user to read aloud. The script should be ~15-20 seconds and include varied intonation:

Suggested script to read aloud (~15 seconds):

  "Welcome to this video walkthrough. Today we'll explore some exciting
   new features — including improved performance, better accessibility,
   and a redesigned dashboard. There's a lot to cover, so let's take it
   step by step and get started!"

Options:
  1. I'll record this script (provide the file path when ready)
  2. I have my own audio sample
  3. Use an existing file (enter path)

Read it at the pace you want the clone to narrate at. The clone inherits its speaking rate from the reference, and temperature/regeneration cannot correct pace afterwards — a short, snappy reference reliably produces rushed (170-210 wpm) narration on every take.

Read the full file on GitHub · 215 lines

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.

  1. 3d ago First seen · 215 lines · 13 tokens per session scan A 02c853057892

Subscribe to this mod's changes

voice-clone is a command published in the GitHub repository digitalsamba/claude-code-video-toolkit (2,035 stars, last pushed 2d ago), licensed MIT. It adds 13 tokens to every session and 1,868 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 2026-08-30.