ElevenLabs

ElevenLabs turns written lines into spoken audio. That description has been true of text-to-speech for twenty years, and the reason this one matters is that the result no longer announces itself as a robot within three words.

What it actually does

You give it text and a voice, and it gives you an audio file. You can do that in a browser for one-off lines, or call it from a script when there are thousands of them.

The second way is where it stops being a novelty. A voice that can be called from code is a voice you can regenerate, which changes what you are willing to edit.

How I use it

It speaks every phrase in the Spanish lessons I make. Each episode is an hour or more of teaching phrases, and generating them is the only reason that is a Tuesday rather than a month of recording.

It also reads the coaching lines in a small boxing app I built for myself. Same job, different clothes.

It does not work alone. The lessons pair it with Kling, which animates the still background the audio sits over, and both are called by a script rather than used by hand. My current setup has the whole arrangement.

Where it fits

The test is whether your script is finished. If it is, recording yourself is often fine and costs nothing.

If your script is going to change, generated speech is the difference between editing a line and re-recording an afternoon. That is the whole argument.

Honest limitations

It bills in credits, so the cost moves with how much you produce rather than sitting still each month. A heavy month is a real bill, and it is worth checking the estimate before you run a long job rather than after.

It is also still a synthetic voice. It is very good, and people who are listening for it will still sometimes hear it. If the warmth of a real human voice is the product, this is not the tool.

Voices, limits and plan names in this corner of the market change often, so check the current pricing page before you budget rather than trusting a figure written down months ago.

Next step

Generate a single paragraph on the free allowance before you commit to anything. If it is going into video, Kling is the other half of the pipeline, and my current setup shows how they fit together.