elevenlabs · elevenlabs · voice & music
ElevenLabs, for everything you'd otherwise record.
Voiceovers, music, sound effects, voice changing, noise removal and transcripts, in one tool next to the video you're putting them on.
- made by
- ElevenLabs
- versions
- 6
- from
- 10 tokens
- included on
- Starter+
best at
What ElevenLabs is good at
ElevenLabs voices handle the parts that usually give synthetic speech away: breaths in the right places, emphasis that follows meaning, a question that actually sounds like one. Eleven v4 is the most expressive of its models, and the delivery presets and sliders let you push a read from calm narration to something more animated.
Speech is only part of it. Sound effects turns a description such as "heavy wooden door creaks open in an empty hall" into a clip of up to 22 seconds. Music writes a track from a description of genre, mood and instruments, from ten seconds to five minutes, with or without vocals. Those cover most of the audio a short video, podcast intro or app needs.
The other three modes work on recordings you already have. Voice changer re-says your recording in another voice, keeping your timing and intonation. Voice isolator strips background noise and music to leave just the voice. Speech to text transcribes audio and can label who is speaking. Any result can be passed straight to another mode, so a noisy interview can be cleaned, then transcribed, in two clicks.
versions
ElevenLabs versions on LazySusan
| Version | Use it for |
|---|---|
| Eleven v4 | Highest quality, most expressive speech. |
| Eleven v4 Turbo | v4 quality with very low latency, at half the price per character. |
| Eleven v3 | Dramatic delivery, more than 70 languages. |
| Multilingual v2 | Natural and stable on long-form reads, 29 languages. |
| Flash v2.5 | Ultra-low latency, 32 languages, lowest price per character. |
| Music, sound effects, voice changer, voice isolator, speech to text | Each a mode in the same tool. |
pricing
What ElevenLabs costs on LazySusan
| Generation | Tokens |
|---|---|
| Speech, Eleven v4, per 1,000 characterssame for v3 and Multilingual v2 | 96 |
| Speech, v4 Turbo or Flash v2.5, per 1,000 characters | 48 |
| Sound effect, any length | 10 |
| Music, 30 seconds | 90 |
| Music, 3 minutes | 540 |
| Voice changer, per minute | 50 |
| Voice isolator, per minute | 50 |
| Speech to text, one hour | 264 |
Speech is priced by character, music by second, sound effects per clip. Uploaded recordings are priced by file size, about 1 MB a minute, with a one-minute minimum. A 900-character script (about a minute of narration) on Eleven v4 is 87 tokens, so Starter's 1,000 monthly tokens go a long way here.
Every plan includes ElevenLabs. Starter is $8 a month with 1,000 tokens; unused tokens roll over while you're subscribed. Compare plans.
prompts
Prompts to try with ElevenLabs
// voiceover
Welcome back. Today we're making the flakiest croissants you've ever tasted, and it's easier than you think.
// phone greeting
Thanks for calling Northwind Dental. Press one to book a visit, or stay on the line to speak with us.
// sound effect
Rain on a tin roof with distant thunder rolling across a valley.
// music
Warm lo-fi hip hop beat for studying, soft piano and vinyl crackle, no vocals, 2 minutes.
// music
Upbeat indie pop jingle for a product launch, hand claps and whistling, 30 seconds.
tips
Getting better results
- 01Write for the ear. Short sentences, commas where you'd breathe and numbers spelled the way they should be said ("twenty twenty-six") make a bigger difference than any slider.
- 02Draft on v4 Turbo or Flash v2.5 at 48 tokens per 1,000 characters, then render the final take on Eleven v4.
- 03Lower stability for a more expressive read, raise it for a steadier one. If a voice starts to sound like a different person, raise similarity.
- 04Clean a noisy recording with Voice isolator before you transcribe it or run it through the voice changer. Both work better on a clean voice.
- 05For sound effects, describe the material and the space, not just the event: "glass bottle rolling on a tiled kitchen floor" beats "bottle sound".
alternatives
When to pick a different tool
- Veo 3.1 if you want sound generated together with the picture, including short lines of dialogue.
- Claude if you need the script written or tightened before it's read aloud.
- ChatGPT if you want quick drafts of ad reads, intros or phone menus to voice.
- Luma if you need silent footage to lay your voiceover or music under.
faq
ElevenLabs on LazySusan: questions
Yes. Text to speech, music, sound effects, voice changer, voice isolator and speech to text are included on every LazySusan plan, from Starter at $8 a month.
96 tokens per 1,000 characters on Eleven v4, v3 and Multilingual v2, and 48 tokens per 1,000 on v4 Turbo and Flash v2.5. About a minute of narration on Eleven v4 is around 87 tokens.
Yes. Describe the genre, mood and instruments and choose a length from 10 seconds to 5 minutes, with or without vocals. A 30-second track is 90 tokens.
Eleven v3 covers more than 70 languages, Flash v2.5 32 and Multilingual v2 29. Speech to text lets you pick the language of the recording.
Yes. Upload a recording in Voice changer and pick a voice. It keeps your timing and intonation, and can remove background noise at the same time.
Yes, subject to ElevenLabs' usage policies.
get started
ElevenLabs and every other model, one plan.
One login, one bill and one Library for everything you make, with the right model for each part of the job.
Starter is $8/month. Cancel anytime.