A Thai line at 0
foithong · paxa-tts-flash-v2
Emotional voices
A steadier read, or a more varied one
Temperature sets how much a line's pacing, pauses, breaths, and melody vary from one request to the next. They vary least at 0 and most at 2.
One line, two takes at each setting#
Foi Thong reads the same line twice at 0, at 0.5, and at 2. Every clip is Paxa TTS Flash v2.
- At 0, take 10:05
- At 0, take 20:05
- At 0.5, the default, take 10:06
- At 0.5, take 20:05
- At 2, take 10:06
- At 2, take 20:05
paxa-tts-flash-v2 · foithong · captured Oct 28, 2026 · unedited
In Thai and Chinese#
One setting reads every language. Each line here is one take at 0 and one at 2.
A Thai line at 2
foithong · paxa-tts-flash-v2
A Chinese line at 0
foithong · paxa-tts-flash-v2
A Chinese line at 2
foithong · paxa-tts-flash-v2
Questions#
How steady is a line at 0?
It varies least of any setting. Two requests for one line in one voice come back closest in timing, pauses, and melody, and two takes can still differ.
Which models read temperature?
Paxa TTS Flash v2. On any other model, only a value that rounds to 0.5 is accepted, and any other is refused before any charge.
How does it work with a designed voice?
As on any other voice. Lower values hold it steadier, and higher values vary it more.
Where do I set it?
In the temperature field of POST /v1/tts, POST /v1/audio/speech, or a live connection's start frame, and with the Temperature slider in the console's Speech tool.
Does temperature cost extra?
No. A request costs the same at every temperature.
Set the temperature. New accounts start with 100 free credits. Move the Temperature slider in the playground and compare two takes.
Start free