tm.sesapi.dev — Turkmen speech & language APIs
================================================

Auth: every request needs   Authorization: Bearer <api key>
(except /docs, /healthz and the /demo page)

ASR  (Speaches, OpenAI-compatible)      base_url = https://tm.sesapi.dev/asr/v1
  POST /asr/v1/audio/transcriptions     multipart: file=<audio>, model=local/whisper-tk, language=tk
  GET  /asr/v1/models

TTS  (f5-tts-tk, OpenAI-compatible)     base_url = https://tm.sesapi.dev/tts/v1
  POST /tts/v1/audio/speech             json: {"model":"all-v5","voice":<voice>,"input":<text>,"response_format":"wav"|"mp3"|"pcm"}
  GET  /tts/v1/voices                   available voices
  GET  /tts/v1/models

LLM  (Ollama)                           base_url = https://tm.sesapi.dev/llm/v1
  POST /llm/v1/chat/completions         OpenAI-compatible, stream=true supported
  GET  /llm/v1/models
  POST /llm/api/chat, /llm/api/generate Ollama native API (also /llm/api/tags)

Demo                                    https://tm.sesapi.dev/demo/   (voice agent, no key)
Playgrounds                             https://tm.sesapi.dev/play/asr/  https://tm.sesapi.dev/play/tts/  (ASR / TTS demo pages, no key)
Health                                  https://tm.sesapi.dev/healthz
Request an API key                      https://forms.gle/CYdLBM5YHsLFoEmT8

Example:
  curl https://tm.sesapi.dev/asr/v1/audio/transcriptions \
    -H "Authorization: Bearer $KEY" \
    -F "file=@audio.wav;type=audio/wav" -F model=local/whisper-tk -F language=tk