Free speech-to-text and text-to-speech API quotas (Oct 2026, with the catches)
Description and use
The author successfully built a prototype using Groq for live transcription, Google TTS for replies, and local Whisper for backlog, while planning to scale with AssemblyAI or Deepgram credits.
Problem
Need voice input and spoken replies for a side project without upfront cost.
Outcome
The author successfully built a prototype using Groq for live transcription, Google TTS for replies, and local Whisper for backlog, while planning to scale with AssemblyAI or Deepgram credits.
Workflow
- 1. Reviewed pricing pages of major speech APIs.
- 2. Recorded free quotas and constraints for each provider.
- 3. Selected Groq for daily‑reset free API for live transcription.
- 4. Chose Google Standard TTS with budget alert for spoken replies.
- 5. Used local Whisper for existing audio backlog.
- 6. Planned to use AssemblyAI or Deepgram credits if project scales.
- 7. Adopted habits: check per‑month/day/one‑time limits, billing enablement, per‑channel/session billing rules.
Evidence
Groq's rate‑limit docs list whisper‑large‑v3 and whisper‑large‑v3‑turbo at 20 requests per minute, 2,000 requests per day, 7,200 audio seconds per hour and 28,800 audio seconds per day.
Speech-to-Text: the V1 API gives the first 60 minutes per month free, then $0.016 per minute with data logging or $0.024 without.
Azure's free F0 tier includes 5 audio hours per month of real‑time speech‑to‑text (shared between standard and custom models, batch transcription not supported) and 0.5 million characters per month of neural text‑to‑speech.
Deepgram gives new accounts $200 in credit with no card required, and says it does not expire.
AssemblyAI gives $50 in free credits, no card. At the listed $0.15 per hour for Universal‑2, that is about 330 hours of pre‑recorded audio.
The Free plan has 10,000 credits per month, shared across text‑to‑speech, speech‑to‑text, music and sound effects.
- Record date: 2026-10-12T09:41:03.090Z