Kiri Voice Recorder

Record a clean supervised set for TTS training. We only need one speaker, one prompt at a time, and crisp read-throughs with no background music, overlap, or improvising.

Speaker ID Session Slug Clip Prefix Target Sample Rate

Prompt List

Microphone not armed

Click “Arm Microphone” once, then record line by line.

Prompts

0

Recorded

0

Ready

0

Uploaded

0

TTS Probe

Quick quality and latency check against the RunPod GPT-SoVITS server. We care about staying under roughly 700ms once the model is warm.

Pod Base URL GPT Weights SoVITS Weights Reference Audio

Prompt Text Target Text

TTS probe idle.

Warmup

--

Headers

--

Total

--

Audio

--

No synthesized clip yet.