The bar at the bottom of the reader is there before any audio exists.

On the far right, the source switcher, reading full, generate.

Click it.

Each source is listed with a status word: ready, generate, a percentage while it runs, or retry.

Under the divider, choose a different voice.

Filter by language, gender and type, or search by name.

Every row previews, one at a time.

Pick one and confirm, and the choice applies to this generation only.

It runs in the open.

The play button counts, and the progress names what it is doing, chunk by chunk, and how many there are.

When it finishes the item plays on its own.

Play, elapsed time, the scrubber whose ticks are the moments in the piece, total time, speed, source.

The words light as they are spoken, and clicking any word sends the voice there.

The player comes with you across Today and Library, so you can keep listening while you move around.

And when the piece ends, it ends.

Nothing queues up behind it.

The essay, read aloud, with the words keeping time.