Hi — we're Hakim (tryhakim.ai), a voice AI platform with our own TTS stack, and we'd like to be considered for the TTS benchmark.
We went through the runner code and methodology docs and think our API maps cleanly onto your existing provider shapes:
POST /v1/audio/speech — buffered or HTTP-streamed synthesis, same shape as your HTTP TTS cohort (OpenAI, ElevenLabs HTTP).WSS /v1/audio/speech/stream — a persistent WS session returning raw PCM-S16LE @ 24kHz mono audio frames, matching your WS streaming cohort (Cartesia, Rime, Hume, Deepgram aura-2).
We're happy to provide a production API key / cover the credits needed to run us, so this doesn't have to sit as PENDING waiting on provisioning. Reach us at engineering@tryhakim.ai and we can get you what you need.
(Opening this per CONTRIBUTING.md's guidance to discuss significant additions via an issue first.)
Hi — we're Hakim (tryhakim.ai), a voice AI platform with our own TTS stack, and we'd like to be considered for the TTS benchmark.
We went through the runner code and methodology docs and think our API maps cleanly onto your existing provider shapes:
POST /v1/audio/speech— buffered or HTTP-streamed synthesis, same shape as your HTTP TTS cohort (OpenAI, ElevenLabs HTTP).WSS /v1/audio/speech/stream— a persistent WS session returning raw PCM-S16LE @ 24kHz mono audio frames, matching your WS streaming cohort (Cartesia, Rime, Hume, Deepgram aura-2).We're happy to provide a production API key / cover the credits needed to run us, so this doesn't have to sit as
PENDINGwaiting on provisioning. Reach us at engineering@tryhakim.ai and we can get you what you need.(Opening this per CONTRIBUTING.md's guidance to discuss significant additions via an issue first.)