Speech generation and the surrounding conversation
ElevenLabs develops generative audio technology, including speech synthesis, voice creation and dubbing. Its company history dates to 2022 and describes improving film dubbing as an early motivation. The product range puts audio generation into both creative production and software that speaks with users. [1]
ElevenAgents adds a conversation layer around the models. Its documentation combines speech recognition, a language model and text-to-speech with interruption handling, turn-taking logic and knowledge bases. Generating an audio file and sustaining an interactive conversation are therefore different parts of the company's contribution. [2]
Direction, dialogue and response time
Eleven v4 and v4 Turbo were released in September 2026. The company emphasized delivery instructions, contextual dialogue between speakers and support for more than 90 languages, making voice direction a more explicit input to speech generation. [3]
Turbo targeted low-latency conversational uses. The company's latency measurements describe its inference tests, rather than the full delay a user experiences after speech recognition, network calls, application logic and audio playback. The release's quality comparisons are likewise reported evaluations, not proof that every generated voice performance will satisfy a particular production brief. [3]