AI audio platform for lifelike text-to-speech, voice cloning, dubbing, transcription, music, and conversational voice agents, available in a web studio and via API.
- Eleven v3 for expressive narration and Flash v2.5 for roughly 75 ms latency
- Library of more than 17,500 voices
- Speech-to-text, dubbing, music, and voice agents on one credit pool
- Paid-plan credits roll over for up to two months
- Free plan has no commercial license
- Professional Voice Cloning starts at the $22 Creator tier
- Shared credits drain quickly when mixing music, dubbing, and agents
ElevenLabs is an AI audio company that grew from a text-to-speech startup into a platform covering speech synthesis, transcription, dubbing, music, and voice agents. One shared credit pool funds every product on a plan, from the web studio to the developer API.
The model lineup includes Eleven v3, which the company presents as its most expressive text-to-speech model, and Eleven Flash v2.5 for low-latency use cases at about 75 ms. Scribe v2 handles speech-to-text, Dubbing v2 carries a speaker’s emotion and delivery across languages, and the Music model produces tracks with vocals or as instrumentals. ElevenAgents configures and monitors conversational agents over phone, chat, email, and WhatsApp. The voice library holds more than 17,500 voices, creative speech and agents cover 90+ languages, and paid plans add Instant Voice Cloning on Starter and Professional Voice Cloning from Creator upward.
Audiobook narrators, YouTube producers, game studios, localization teams, and developers building voice assistants make up the core audience. The credit model is the main thing to watch: generous-looking allowances drain quickly when the same pool funds music, dubbing, and agents, and the free plan carries no commercial license. Unused paid credits roll over for up to two months, which softens the cost of uneven monthly usage but does not eliminate it, and high-volume API work usually lands on the Pro tier or above for 44.1 kHz PCM output.
