Offline voice packs for apps and devices
Checked phrases as audio files you buy once and play from your own app, kiosk or device.
What you get
- Audio: ready-generated M4A recordings, one file per phrase.
- Data:
manifest.jsonandphrases.csvmapping every file to its text. - Licence:
LICENSE.txtin every pack.
Available now
253 packs (en: 50, es: 37, fr: 21, hi: 48, ja: 46, pt: 32, zh: 19), one-time $24 to $199 each. Browse and hear samples on the packs page.
Pack or API?
| Offline pack | API | |
|---|---|---|
| Pricing | One time | $3 per million characters |
| Speed | Instant, plays locally | About 0.14 s stored, 0.4 to 0.8 s new on a paid account |
| Needs network | No | Yes |
| Text | Fixed phrases | Any text |
Questions
What are offline voice packs?
One-time purchases of ready-generated recordings of checked phrases (menus, confirmations, errors, greetings) as M4A files with a manifest.json and phrases.csv. They play from your own app or device with no per-play fee and no generation delay.
What can I do with the files?
Play them inside your own apps, devices and services, offline, commercial use included. You may not resell or redistribute the files or use them to train speech models. The terms are in each pack's LICENSE.txt.
How much do they cost?
Today: 253 packs (en: 50, es: 37, fr: 21, hi: 48, ja: 46, pt: 32, zh: 19), one-time $24 to $199 each. There is no subscription.
Can I hear one first?
Yes. Every pack has a free 25-phrase sample at /packs/{id}/sample.zip.
Why use a pack instead of the API?
Prompts that never change (kiosk, IVR, device messages) cost nothing to play after the one-time purchase, and they work without a network connection. Safety-critical alerts should not depend on the network. Use the API for text that changes.
Are all languages human-reviewed?
No. Non-English clips are marked native_review_pending until a native speaker has reviewed them, and Hindi is a preview. Voices are AI-generated, not human recordings.