# PlayHT streaming caps: 20k chars per request, PlayDialog for two speakers

## TL;DR

When building on PlayHT streaming: chunk text at 20,000 characters per request, the API will not accept more. For two-speaker audio use PlayDialog with voice and voice_2 and a turn_prefix, do not try to fake it with one voice. If you need a brand-new designed voice there is no public endpoint for that, your options are cloning or the prebuilt library.

Use this when you run into the situation in the title.

**When not to use this skill:** unrelated tasks. It covers only the procedure above.

## Compatibility

The steps above apply to the commands named in them. This skill does not pin a version, so if a flag looks different on your machine, check your installed version's docs first.

## Details

And pick the model by latency budget, 3.0 Mini answers faster than 2.0.

Context: Web (dsh-voice-mode competitive TTS writeup, citing PlayHT docs): documents real API caps. Play 3.0 Mini caps streaming requests at 20,000 characters per request, longer text must be chunked client-side. Multi-speaker dialogue works via PlayDialog with voice and voice_2 plus a turn_prefix. Voice design (text to brand-new voice) has no public endpoint, only cloning and prebuilt voices. Streaming first-packet latency differs by model: Play 2.0 around 230 ms, Play 3.0 Mini around 190 ms.
