ElevenLabs Launches Speech Engine to Turn Chat Agents Into Voice Agents With One Prompt.

ElevenLabs has introduced Speech Engine, a new capability that allows developers to convert existing text-based chat agents into full voice agents using a single prompt. The feature integrates speech, transcription, and voice orchestration into one pipeline that layers on top of a developer's current infrastructure without requiring any rearchitecting of the underlying text agent.

The company announced on LinkedIn that Speech Engine is available now via ElevenAPI, positioning voice as the fastest and richest way to exchange information in customer-facing products.

Speech Engine combines ElevenLabs' speech synthesis, real-time transcription, and voice orchestration models into a unified pipeline. Developers can integrate it directly with their existing text-based conversational agents without modifying the core agent logic or infrastructure. The original text-based system remains intact, and the voice layer sits on top, handling audio input, transcription, text processing through the existing agent, and speech output.

This approach contrasts with building a voice agent from scratch or replacing an existing system entirely. Instead, developers add voice capability as a new interface to an agent they have already tuned and tested for text interactions.

Speech Engine is accessible via the ElevenAPI, ElevenLabs' developer-facing product line. The API unifies voice synthesis, transcription, and orchestration in a single call, reducing the complexity of managing multiple vendors or services for different parts of the voice pipeline.

Developers can configure Speech Engine with a prompt that defines the agent's behavior, voice characteristics, and conversational style. The system then handles incoming audio, transcribes it, routes the text through the existing chat agent logic, and returns synthesized speech in real time.

The integration model means teams do not need to rewrite agent logic, retrain models, or migrate data. They connect Speech Engine to their text-based agent endpoint and configure voice parameters. This lowers the barrier for companies already running chat agents on platforms like Voiceflow, custom Bubble backends, or internal APIs.

ElevenLabs has been expanding its voice and audio offerings aggressively. The company recently added multimodal support to ElevenAgents across WhatsApp and web, and cut API pricing up to 55 percent while introducing a pay-as-you-go model.

Speech Engine extends the company's reach into conversational AI by targeting the large base of existing text agents in production. Many enterprises and startups have invested in building and tuning chat agents over the past few years. Speech Engine gives them a path to voice without starting over.

The announcement follows the company's Series D fundraise in February, which brought ElevenLabs to an $11 billion valuation, and a recent milestone of crossing $500 million in annual recurring revenue. The company has also been expanding geographically, including new presences in Australia, New Zealand, and Spain.

Developers can access Speech Engine through the official ElevenLabs API documentation and sign up for an API key to start testing.