Voice Buddy

Voice Buddy turns a script into a finished read. It generates speech, clones a voice from a single clip, converts one performance into another voice, transcribes, dubs into other languages and narrates long-form work, from script to finished, mixed audio.

How it works

1. Choose or clone a voice. Start from the voices that ship with it (200 of them), or upload one clean clip of a voice and use it the moment the upload finishes.
2. Write, upload or point at a file. Type a script, drop in a recording to re-voice, or send text through the API from your own systems.
3. Ship it. Download the master, export a whole project, or have a webhook tell your system the render has landed.

Masters are 44.1 kHz WAV with no watermark.

What it does

Speech from a script. Type or paste copy and get a finished read in the voice you chose. A pause marker gives you exactly the silence you ask for, to a tenth of a second.

Voice cloning. One clean clip is enough to clone a voice. A clone works as soon as it uploads.

Voice conversion. Convert a recorded performance into another voice.

Transcription and dubbing. Transcribe recordings, or send an episode, audio or video, and get it back re-voiced in another language.

Long-form projects. A book or long script is split into chapters and paragraphs, each rendered on its own, so fixing a typo in chapter forty re-reads one paragraph. Projects render on a queue, so you can close the tab.

Copy written to the slot. Turn an email, a web page or a document into copy cut to a real word budget for the length you need, such as 72 to 84 words for a :30.

Batch reads. Put a tag in the copy, paste rows from a spreadsheet, and get one take per row, up to 500, priced before you commit.

Editing a take

• Fix one word. Click the word that came out wrong, retype it, and it re-renders in the same voice and splices back in. The rest of the take is untouched.
• Say it properly. Flag a mangled word and say it into your microphone, and it comes back in the voice that got it wrong. That fix does not count against your quota.
• Names said right. Write a respelling once, and every take, API call, project and scheduled bulletin uses it.
• Free edits. Cuts, fades and gain changes cost nothing and can always be undone. Undoing a re-rendered word does not bill you again, and saving writes a new clip so the original survives.
• Pin the take. Every take shows the seed it came from. Pin the one the client approved and it comes back identically on a later render.
• Rescue a bad recording. Four purpose-built clean-up chains, including one for interviews that keeps the pauses. Clean-up costs no characters.

For broadcasters

Voice Buddy can write, voice and deliver traffic and weather bulletins on a schedule, whether or not anyone is in the building.

• Draw the area your signal covers and name the roads in and out of it. Bulletins rank live incidents down to the few worth airing.
• An empty incident list gets the clear-roads line you wrote. It never invents an incident.
• Weather bulletins lead on an active warning rather than the weekend outlook, in your market's units and clock.
• Any voice in the workspace can read them, including a clone of the presenter who would have read them live.
• Voice and music bed are mixed through a broadcast processing chain. Mixing costs no characters.
• Bulletins are delivered by pull URL, webhook, FTP, SFTP or a cloud folder, either overwriting one file name for a watch folder or stamped individually for an archive.

What it will not do

Voice Buddy does not offer laugh or sigh tags. They were measured on this engine and they do not work, so they are not offered.

Billing

Voice Buddy is billed on the characters of speech you generate, with a free plan of 10,000 characters a month. Current plans and prices are on the Voice Buddy site.

• Quota is reserved when a render starts and refunded if it fails, so you pay only for finished work.
• Transcription is billed at a tenth of the synthesis rate.
• Editing, mixing and clean-up cost nothing.

Why it works this way

We run the speech engine ourselves rather than reselling another company's. That is why a clone is usable the moment it uploads, why nobody takes a cut of every render, and why the prices and rate limits are ours to set. See How we build.

For developers

The REST API covers everything the product does. The reference, code samples, playground and OpenAPI document are generated from one specification, so they cannot disagree, and one key drives both the REST API and AI-assistant access. An agent can also drive the API on your behalf at your own role, and asks you to confirm anything destructive. See the developer section of the Voice Buddy site.