On air
Now playing
Booth v3
///
New
32 languages
///
Latency
300 ms to first audio
///
Studio
Dubbing for video, now in beta
///
Product
Four channels.
One booth.
Text to speech, voice cloning, dubbing and live agents run on one engine, one voice library and one API key. Bring a script, leave with a master.
What each channel does
01
Built for the take, not the demo. Every channel ships with the controls a producer would ask for, and none of the ones they would not.
CH 01 / Text to speech
Direct the read like a session.
Set stability, clarity and style per line. Add pauses with a tap, fix one word without regenerating the paragraph.
Max script
50,000 chars
Output
MP3 / WAV / PCM
Sample rate
Up to 48 kHz
Speech
Clone
Dub
Agents
On air
Good evening. You are listening to the late edition, read by a voice that was never in the room.
Voice
Mara
Stability
62
Clarity
80
Style
35
CH 02 / Voice cloning
One minute in. A voice out.
Record in a quiet room, read the consent line, and the clone lands in your library ready for every other channel.
Input
60 sec of audio
Consent
Spoken + verified
Ready in
Under 3 min
New voice / Ilja
Recording
00:47
of 01:00 needed
Step 01
Record
One quiet minute
Step 02
Verify
Read the consent line
Step 03
Ship
Live in the library
CH 03 / Dubbing
Same voice. Every market.
Upload a video and Booth splits speakers, translates the script and fits each line to the original timing.
Languages
32
Speakers
Up to 9 per file
Formats
MP4 / MOV / SRT
Scene 12 / Shot 03
Timing locked
EN
We open at nine. Bring the whole team.
DE
Wir öffnen um neun. Bring das ganze Team mit.
JA
9時に開店します。チーム全員でどうぞ。
ES
Abrimos a las nueve. Trae a todo el equipo.
PT
Abrimos às nove. Traga a equipe toda.
CH 04 / Voice agents
A voice that picks up.
Connect your knowledge base and phone line. Agents handle interruptions, look things up and pass to a human with notes.
Latency
300 ms first audio
Channels
Phone / Web / App
Handoff
Warm transfer
Line 04 / Support
Live call
Caller
00:12
Hi, my order says delivered but there is nothing at the door.
Booth agent
280 ms
Sorry about that. I can see it was left at the side entrance at 2:14. Want me to send a photo?
Caller
00:21
Oh. Yes please.
Booth agent
240 ms
Sent to your phone. Anything else?
Avg response 262 ms
Developers
02
Ship it in an afternoon. One REST call for a file, a WebSocket for streams, and SDKs for the stacks you already run.
cURL
Python
Node
POST /v3/speech
# Generate speech in one callcurl https://api.booth.audio/v3/speech \-H "Authorization: Bearer $BOOTH_KEY" \-d '{"voice": "mara","text": "Good evening. This is the late edition.","language": "en","stream": true}' \--output take-04.mp3
200 OK / first audio in 284 ms
Python
pip install booth
Node
npm i @booth/sdk
Go
go get booth
Swift
SPM package
Kotlin
Maven Central
Webhooks
Signed events
300 ms
First audio
32
Languages
99.99%
Uptime / 90 days
Booth / Take 05 / Your turn
Press
record.
Ten thousand free characters a month. No card, no call, no waiting list.