Hi LiveKit community 👋
I’d like to share an open-source tool I built called
Hikaku-Voice.
https://github.com/morix1500/hikaku-voice
Hikaku-Voice is a simple tool to
compare STT and TTS models side by side, focusing on:
• Transcription / speech quality
• Latency (response time, TTFB for TTS)
• Real-world usability for voice AI applications
I built this because keeping up with new STT/TTS models is getting really hard, and validating quality + latency usually requires engineering effort. I wanted something that
engineers and non-engineers (PMs, designers) can use easily.
The core of this tool is powered by
LiveKit Agents plugins.
Thanks to the plugin architecture, adding or swapping STT/TTS providers is very straightforward, and you can even compare multiple models from the same provider at once.
If you’re building voice AI apps or experimenting with STT/TTS providers, I hope this is useful.
Feedback and contributions are very welcome 🙏