ElevenLabs is an AI voice company; its products are built on its own audio research. Its main product is text-to-speech (TTS), the conversion of written text into spoken audio; it also offers speech-to-text, voice cloning, dubbing into other languages, music and sound effects.
Each voice has an ID that a program passes in an API request. The voice library holds more than 10,000 voices, and a user can clone a voice from an audio recording or design one from a text description. Models differ in quality, speed and language coverage: Eleven v4 gives the most expressive speech in more than 90 languages, and the fastest models have a model inference latency of about 75 to 100 milliseconds. ElevenLabs is used in four ways: ElevenCreative, a web editor for voiceovers, dubs and music; ElevenAgents, for voice agents that talk with callers; ElevenAPI, a REST interface with Python and TypeScript SDKs; and Reception.ai, a phone receptionist.
Plans are monthly subscriptions with credits; text-to-speech costs one credit per character. A free plan includes 10,000 credits a month, and API users can instead prepay a balance that is drawn down with use.