Compare
The cheapest speech-to-text API, compared
Published prices per audio hour for recorded and live transcription, side by side. Giggy transcribes recordings for $0.04 an hour on Fast and live audio for $0.08 an hour on Streaming.
Speech to text: what you pay
| Provider | Mode or model | USD per audio hour |
|---|---|---|
| Giggy | Fast | $0.040 |
| AssemblyAI | Universal-2 | $0.150 |
| OpenAI | gpt-4o-mini-transcribe | $0.180 |
| Google Cloud | Speech-to-Text V2 · Dynamic batch | $0.180 |
| Azure | Speech · Batch transcription | $0.180 |
| ElevenLabs | Scribe v3 | $0.200 |
| AssemblyAI | Universal-3.5 Pro | $0.210 |
| Deepgram | Nova-3 · Pre-recorded | $0.258 |
| Deepgram | Nova-3 Multilingual · Pre-recorded | $0.312 |
| OpenAI | Whisper / gpt-4o-transcribe | $0.360 |
| Amazon | Transcribe · Standard batch | $0.360 |
| Provider | Mode or model | USD per audio hour |
|---|---|---|
| Giggy | Streaming | $0.080 |
| Soniox | stt-rt-v5 · Streaming (est.) | $0.120 |
| ElevenLabs | Scribe v2 Realtime | $0.220 |
| Speechmatics | Linden-1 · Agent STT | $0.300 |
| OpenAI | GPT-4o Transcribe (est.) | $0.360 |
| Speechmatics | Real-time Enhanced | $0.430 |
| AssemblyAI | Universal-3.6 Pro Realtime | $0.450 |
| Deepgram | Nova-3 · English streaming | $0.462 |
| Deepgram | Flux Multilingual · Streaming | $0.468 |
| Google Cloud | Gemini 3.5 Transcribe Live (est.) | $0.536 |
| Amazon | Transcribe · Standard Streaming | $0.600 |
| Google Cloud | Chirp 3 · Streaming | $0.960 |
| Azure | Speech · Standard realtime | $1.00 |
| OpenAI | GPT-Live-Transcribe | $1.02 |
All recorded-audio prices checked November 11, 2026 on each provider's pricing page or official price API. ElevenLabs, OpenAI, Deepgram, Google, Amazon, Azure and AssemblyAI checked November 11, 2026; Speechmatics October 8; Soniox token-based estimate October 20.
Recorded audio and live audio
Most providers price recorded files and real-time streams differently. Recorded files are compared with Giggy Fast, which accepts recordings up to 12 hours. Live audio is compared with Giggy Streaming over a WebSocket.
Each price is the provider's published pay-as-you-go rate, converted to US dollars per audio hour and linked to its source. Add-ons such as speaker diarization, volume tiers and promotions are excluded.
What the difference means at scale
Transcribing 1,000 hours of recordings costs $40 with Giggy Fast. The same hours cost about $150 on AssemblyAI Universal-2, $180 on OpenAI gpt-4o-mini-transcribe or Google dynamic batch, $258 on Deepgram Nova-3 and $360 on OpenAI Whisper or Amazon Transcribe.
Accuracy matters too
A cheap transcript is only useful if it is accurate. Giggy publishes word error rates on FLEURS across 12 languages and on the Pipecat English benchmark, including the languages where other models score better.
Frequently asked questions
What is the cheapest speech-to-text API?
Among the providers compared here, Giggy has the lowest published paid rates: $0.04 per hour for recorded audio on Fast and $0.08 per hour for live audio on Streaming. Giggy Batch is free for recordings up to 25 minutes.
Is real-time transcription more expensive?
Usually. Most providers charge more for live streaming than for recorded files. Giggy Streaming costs $0.08 per audio hour, still below every live rate compared here.
How is transcription billed?
Giggy charges the validated duration of the input audio and rounds up to one credit per request. Prices here exclude add-ons such as speaker diarization.