A voice for the end of the day
Guided meditation, dhamma talks, and sleep audio ask a voice to do the opposite of everything else on this site. Three of the catalog's voices were cast for exactly that.
3 voices, one line#
Each was cast for a different kind of quiet, and each reads the same line below. Listen with the volume where a listener would actually have it.
The line
หายใจเข้าช้า ๆ แล้วค่อย ๆ ปล่อยออกมา ไม่ต้องรีบไปไหนทั้งนั้น
One line, every calm voice. Names link to each voice's own page.
Most speech work wants to be heard. This work wants to be almost not noticed: a voice close to the ear, unhurried, with nothing in it that asks the listener to sit up. A voice that performs is wrong here in a way that is obvious in the first sentence and hard to describe afterwards.
Thai audio of this kind is made constantly and served narrowly. A meditation app records a teacher for a season and then cannot add a session without booking them again. A temple broadcast reads the same passage every morning in whoever is available. Both are the same problem: the work is continuous and the voice is a booking.
What gets read this way#
Four kinds of audio that sound like each other and are made by people who rarely meet.
Guided meditation
A session is a script with pauses in it. Once the script is a file, a library grows by writing. A course can ship its second week in the week after its first.
Dhamma talks and readings
Passages that are read the same way every time, often daily, often by whoever is free that morning. A fixed voice reads them the same way on the hundredth morning as on the first.
Sleep and wind-down audio
Long, undemanding, and listened to by someone who intends to stop paying attention. Length is cheap here: an hour of speech is priced by its characters like anything else.
Waiting rooms and quiet spaces
Clinics, spas, and retreat centres play the same few minutes on a loop. Changing what it says is an edit, and the voice on the loop does not have to be re-booked to change one line.
Pace is part of the register#
The same words land differently at a different speed, and this is the one kind of audio where slow is the point. A guided session at full pace reads as instruction; the same script slowed reads as company.
Every request takes a rate as a multiplier on the voice's own speed. It changes how the audio sounds and never the character count. A slower read costs exactly what the faster one did.
Rate range
0.5–1.5×
The rate is a per-request field. A session can open at one pace and close at another without changing voice.
Questions#
- Can the voice pause between instructions?
- Pauses belong to the script and to the player. Send each passage as its own request and lay the silences between them in your timeline, which also lets you change a pause without re-reading the passage.
- How slow can it go?
- Down to 0.5 times the voice's own speed, and up to 1.5. The charge is the same at every rate, because it follows the characters and not the seconds.
- Will the voice sound the same next season?
- Yes. A voice is a fixed id in the catalog. A library recorded across two years is one voice throughout, with nobody to bring back.
- Can we publish this audio, or sell it in an app?
- Output made on free credits is for personal use and evaluation. Once you buy a plan or credits, generated output becomes yours for any use, including publication inside a paid app, and that covers output generated before the purchase.
- Do you make any claim about what this does for a listener?
- No. This page is about how the audio is produced. What the listening does for someone is between them and whoever wrote the words.
Free credits, and no card
Sign in and the free grant covers a full session script and the two slower versions you will want to compare it against.