Multi-speaker Dialogue Synthesis: Full Parameter Example

Using the ElevenLabs multi-speaker dialogue synthesis pipeline as an example, this demonstrates how to organize a role-based dialogue array and global synthesis parameters in a single submission. In base, dialogue is a required field; each item contains text and voice (voice takes a preset voice_id, which can be compared one by one at the static preview address). The total character count of all lines is subject to a 5000 upper limit; stability takes an enum of three levels: 0 / 0.5 / 1, used to trade off between "emotional variation" and "timbre consistency"; language_code is an ISO language code enum, and if left empty, the model detects it automatically. This piece constructs three short dialogue segments: a calm narrator, a questioner with a rising tone, and a steady answerer. By switching voice_id and stability levels within the same request, it presents how multi-speaker dialogue synthesis is organized in a single call. It is suitable for developers to quickly understand the input structure, parameter semantics, and multi-role orchestration approach of this pipeline, and can also serve as a prototype starting point for scenarios such as customer service voice agents, podcast conversations, and instructional audio.