GCP Speech-to-Text cost: per minute, by model and features
Speech-to-Text bills per minute of audio (rounded up), with higher rates for premium and specialized models and features like diarization. Long audio and premium models drive the bill. Here is the per-minute model and how to control it.
Quick answer
Speech-to-Text bills per minute of audio (rounded up per request), with higher rates for premium and specialized models (medical, telephony, enhanced) and features like speaker diarization, with a free tier and volume discounts. Cost scales with audio duration, so transcribing only what you need, using the standard model where it suffices, and enabling features selectively are the levers.
Google Cloud Speech-to-Text transcribes audio, billed per minute processed. The rate depends on the model and features you choose, so the cost is a function of how much audio you transcribe and how premium a model you use.
Per minute, by model
| Model / feature | Cost |
|---|---|
| Standard model | Base per-minute rate |
| Enhanced / specialized models | Higher per-minute rate |
| Diarization, punctuation, extras | May add to the rate |
The standard model bills a base per-minute rate, rounded up per request. Enhanced and specialized models (tuned for telephony, medical, or specific domains) cost more per minute, and features like speaker diarization can add to the cost. A free tier covers initial usage, and rates fall at volume.
What drives the bill
Audio duration is the driver: transcribing a large library of long recordings costs in proportion to its hours. Choosing a premium model where the standard would suffice, and enabling features the use case does not need, multiply the per-minute rate. Transcribing silence or irrelevant audio wastes minutes.
Controlling Speech-to-Text cost
Transcribe only the audio you need, trimming silence and irrelevant segments, use the standard model unless a specialized one is genuinely required, enable features like diarization only where needed, and cache transcripts so you do not re-transcribe. The duration-and-model levers mirror AWS Transcribe: send less audio and use the right model.
FAQ
How is GCP Speech-to-Text priced?
Per minute of audio processed, rounded up per request, at a base rate for the standard model and higher rates for enhanced and specialized models (telephony, medical, domain-tuned), with features like speaker diarization potentially adding to the rate. A free tier covers initial usage and rates fall at volume.
How do I reduce Speech-to-Text cost?
Transcribe only the audio you need by trimming silence and irrelevant segments, use the standard model unless a specialized one is genuinely required, enable features like diarization only where needed, and cache transcripts to avoid re-transcribing the same audio. Duration and model choice are the main levers.
Why do enhanced Speech-to-Text models cost more?
Because they are tuned for specific conditions, such as telephony audio or medical terminology, and deliver higher accuracy in those domains, priced at a higher per-minute rate. Use them only where the audio type needs their specialized recognition; for general audio, the standard model is cheaper and sufficient.
What drives Speech-to-Text cost?
Audio duration, since billing is per minute. A large library of long recordings costs in proportion to its hours. Choosing premium models where standard would suffice and enabling unneeded features multiply the per-minute rate, and transcribing silence or irrelevant audio wastes minutes.
Does speaker diarization cost extra?
It can add to the per-minute cost, depending on the model and configuration, since identifying and separating speakers is additional processing. Enable diarization only where the use case genuinely needs to distinguish speakers, rather than by default, to avoid inflating the rate.
Does C3X estimate Speech-to-Text cost?
Speech-to-Text cost is driven by audio duration and model choice, which are usage inputs. C3X prices the surrounding infrastructure, and you model your audio volume and the model and features needed to estimate the per-minute transcription charges.
What to do next
Estimate the infrastructure around your speech pipeline before you build it. C3X reads your Terraform and prices your resources against a live catalog. Start with the quickstart.
Related posts
- GCP Cloud Build cost: you pay per build-minute
Cloud Build bills per build-minute by machine type, with a daily free allowance. Slow builds and oversized machines drive the cost. Caching, parallelism, and right-sized machines are the levers. Here is how the per-minute model works.
- GCP Document AI cost: per page, by processor type
Document AI bills per page processed, with rates by processor, cheap general OCR, more for form and specialized parsers, and higher for custom and generative processors. Running specialized processors on simple documents over-pays. Here is the model.
- GCP Vision AI cost: per image, by feature
Cloud Vision bills per image per feature (label, face, text, object, safe-search, and more), so requesting several features on one image bills for each, with a free tier and volume discounts. Analyzing more features than needed drives the bill.
- GCP Secret Manager cost: per secret version and per access
Secret Manager bills about $0.06 per active secret version per location per month and about $0.03 per 10,000 access operations. It is cheap unless you store many versions or access secrets on every request. Here is the model and the caching lever.
Share this post
Try C3X on your own Terraform
Free and open source. No API key required. One command to install, one command to estimate.