Three ways to meditate
In silence, with a timer. You choose how long. A bell sounds at the start, bells can sound at intervals if you want, and another rings at the end. You can also add settling time before the count begins.
With music. A continuous sound bed, with no voice: a low drone, distant bowls, a slow chord, rain, a stream, wind, the sea, the night or brown noise.
Guided. A voice, or the text on screen, tells you what to do with your attention, with silences between one instruction and the next. You can also put music underneath.
Bells and music are made as they play
The bells (a singing bowl, a large bell, a small chime and a wood block) and the sound beds are not recordings. The browser synthesizes them as they play, using the Web Audio API. According to MDN's documentation, that API works with a graph of audio nodes linked together, and some nodes, like oscillators, generate sound by computing it mathematically.
That has two practical consequences. The sounds do not loop like a file starting over, and they work offline once the page has loaded. And one honest limitation: a synthesized sound does not sound like a real bowl. It is an approximation meant to mark and accompany, not to imitate an instrument.
A script with silences
A guided meditation is a script: a list of short lines, and after each one, a silence. At the start the silences are short and fixed. In the central part of the practice, some silences are elastic: they grow when you choose a longer session. That way one script serves several lengths, and in the longest each silence is still introduced by a line that says what to do in it.
The full script can be read on each meditation's page, before you begin or instead of listening. If you would rather read than listen, you lose nothing.
The voice: today, your device's own
Today the lines are read by the speech voice of your device or browser. That voice changes from one device to another: on a phone, on a computer or in a different browser it may sound different, more natural or more robotic. Meditate does not control how it sounds.
If your device has no voice for the meditation's language, the lines are shown on the screen, at the script's pace. You can also ask for the lines to be shown while the voice speaks.
A screen made for closed eyes
During a session, the controls dim after a few seconds so they do not distract you, and a tap on the screen brings them back. You can keep the screen on for the length of the session, so your phone does not lock.
At the end you can save the session to your practice, with how you feel and a note if you like. It is not required.
Your own sessions
You can build your own timers, with the bells and times you prefer, and write your own guided meditations with your lines and your silences. To share one, a link is generated that carries it with it. How all of that is stored, and what leaves your device, is in the article on privacy.
Common questions
Are the bells recordings?
No. They are synthesized in your browser each time they play, so there is no audio file behind them. That is why they work offline and do not loop.
Why does the voice sound different on my phone and my computer?
Because the voice is the one your device or browser provides, and each has its own. If there is no voice for the language, the lines appear on the screen.
How can one meditation last different lengths?
Some silences in the script are elastic and stretch when you choose a longer session. The lines are the same; what changes is the time you have to practice between them.
Do I have to look at the screen?
No. It is made for closed eyes: the controls dim and the voice carries the session. A tap brings them back.