Create with your own AI.

Give your AI a file, an idea and a credit budget. The HanoLab skill helps it generate music, clone and convert voices, separate stems, master audio and download results.

Prefer an account connection? Connect your AI through MCP to upload audio and convert it with a saved voice. A skill link alone does not install tools or grant terminal access.

Install and connect

npm install -g https://hanolab.com/downloads/hanolab-cli-0.2.1.tgz
hanolab auth login
hanolab skill install

Requires Node.js 20+, local terminal and file access, and a HanoLab API key (Creator or Pro). Create a key on your account page, then enter it in the CLI’s hidden prompt. Keep it out of chat and project files. Use hanolab skill install --path DIRECTORY to choose a skill folder, then start a new assistant session.

Tell your AI what to make

Try: “Use $hanolab to convert this song with my saved voice, using at most 4 credits, and save the audio here.” Or ask it to create a voice from your reference recording, separate a song’s stems, or generate an instrumental.

What you can do

Music generation supports Draft, Studio, Signature and Maestro. Cloning supports Flash and Pro; Pro accepts additional --training-file inputs. Separation supports 2, 4 or 6 stems. Mastering requires a reference track. Credits, voices, uploads, estimates, task status and downloads are also available.

hanolab credits
hanolab voices
hanolab estimate --file /absolute/song.wav
hanolab convert --file /absolute/song.wav --voice VOICE_ID \
  --max-credits 4 --full-song --wait --output /absolute/result.wav

hanolab estimate --operation generate --tier studio
hanolab generate --prompt "Warm acoustic guitar, instrumental" \
  --tier studio --max-credits 4 --wait --output /absolute/music.mp3

hanolab estimate --operation clone --kind flash
hanolab clone --file /absolute/my-voice.wav --name "My voice" \
  --kind flash --consent --max-credits 10 --wait

hanolab separate --file /absolute/song.wav --mode 2stem --max-credits 2 --wait
hanolab download --task TASK_ID --index 0 --output /absolute/stem.wav

hanolab master --file /absolute/mix.wav --reference /absolute/reference.wav \
  --max-credits 1 --wait --output /absolute/master.wav

Use exact voice IDs. Clone only voices you own or have permission to use. No lyrics means instrumental; use --lyrics-file for a song with lyrics (Maestro is instrumental only). MP4 input produces audio. Estimate before submitting and set --max-credits to your budget. Run hanolab help for all options.

Continue after interruption

hanolab resume --job /absolute/receipt.json --wait

Every paid command saves a private receipt before submission. Resume it to continue the original task without creating another paid task. Add --output for an audio result. Cloning returns a voice model ID. Downloads refuse to overwrite existing files; a failed task needs a separate decision before a new paid retry.