Speech-to-text API pricing, per minute

Per-minute list prices from the official pages, converted only where the provider quotes hours. Read 2026-10-08. Accuracy is not part of this table.

Per-minute price, cheapest first

ProviderModelUSD per minuteSource
Speakvoraspeech to text$0.0015this site
ElevenLabsScribe v2 ($0.22 per hour)$0.0037source
DeepgramNova-3 monolingual pre-recorded (promotional)$0.0043source
DeepgramNova-3 monolingual streaming (promotional)$0.0048source
OpenAIwhisper$0.0060source
ElevenLabsScribe v2 Realtime ($0.39 per hour)$0.0065source
CartesiaInk on Scale ($0.39 per hour)$0.0065source
DeepgramNova-3 monolingual, regular price listed$0.0077source

Deepgram shows its Nova-3 prices as promotional with a higher regular price ($0.0077 monolingual, $0.0092 multilingual on the page read). Cartesia's own unit is 3 credits per second; the page gives $0.39 per hour on Scale.

Monthly cost for three workloads

WorkloadMinutes a monthSpeakvoraDeepgram Nova-3 (promo)OpenAI whisperElevenLabs Scribe v2
Voicemail transcription5,000$7.50$21.50$30.00$18.33
Call recording20,000$30.00$86.00$120$73.33
Meeting archive100,000$150$430$600$367

Method: minutes × per-minute list price. Free allowances and minimums are ignored. Speakvora bills per second of audio; the others may round differently.

What the price does not tell you

  • Accuracy: we have not published a benchmark and neither does this page. Test on 100 of your own recordings.
  • Features: Speakvora has no word timestamps or speaker labels yet and takes up to 4.5 MB per request; send longer audio in pieces.
  • Live transcription: Speakvora offers a WebSocket with automatic end-of-speech detection. In our 2026-10-08 production run a 4 second clip took 0.67 s and a 15 second clip 0.90 s (method).

More: · · · how our transcription works