Voice API
voice·api.spareapi.ai·Voice
First-party voice relay served directly by the SpareAPI gateway (no operator needed). ASR: upload audio via a presigned URL, then transcribe with word-level timestamps and speaker diarization — auto-routed across ElevenLabs Scribe (90+ languages) and Aliyun DashScope (fun-asr / paraformer, best for zh). TTS: synthesize MP3 with Microsoft Edge voices (free), ElevenLabs, or CosyVoice. Billing is metered against your USD balance: ASR by audio duration, TTS by character count — the exact cost comes back on every response.
Uptime
99.5%
P95 Latency
6.5s
Total Calls
1.2k
Billing
Per call
/tts/synthesize
Text → MP3. Providers: edge (free), elevenlabs ($0.15/1k chars), cosyvoice ($0.0105/1k chars). Metered by characters; cost returned in X-Cost-Usd.
Request · cURL
curl -X POST https://api.spareapi.ai/tts/synthesize \
-H "xapi-key: ${SPAREAPI_KEY}" \
-H "Content-Type: application/json" \
-d '{
"text": "SpareAPI metered billing test, should debit now.",
"prefer_provider": "elevenlabs",
"voice": "21m00Tcm4TlvDq8ikWAM"
}'Hits /api/scrapeon this origin. Currently returns the endpoint's example response with simulated latency — wire the upstream in /api/scrape/route.ts.
Example response · 200
SUCCESS"(binary audio/mpeg) + headers: X-Provider: elevenlabs · X-Char-Count: 48 · X-Cost-Usd: 0.007200 · X-Balance-After-Usd: 4.992800"