How much does voice transcription cost per minute?

Managed ASR APIs typically charge between $0.004 and $0.02 per minute of audio. Real-time streaming often costs 1.5-2x batch pricing. Add-on features like diarization, PII redaction, or custom models can increase cost.

Which ASR provider is cheapest for real-time voice agents?

Deepgram Nova 2 and AWS Transcribe are usually the cheapest for real-time transcription at scale, while OpenAI Whisper API is simple and low-cost for batch uploads but lacks streaming.

Should I pay for diarization separately?

If you need to know who spoke when (meetings, sales calls, support QA), diarization is essential. Several providers include it free or at low incremental cost; others charge a separate per-minute fee.

🎙️

Voice-to-Text Cost Comparator

Estimate monthly ASR spend across major transcription APIs, then pick the cheapest provider for your voice-agent or meeting-bot workload.

Audio minutes / month

Batch %

Diarization %

Scenario preset

Provider	Model	Effective $/min	Monthly cost
OpenAI Whisper API	whisper-1	—	—
Deepgram Nova 2	Nova-2	—	—
AssemblyAI Universal	Universal-1	—	—
Google Cloud Speech-to-Text	enhanced / latest	—	—
Azure Speech Services	Fast/Custom	—	—
Amazon Transcribe	standard / medical	—	—
Rev AI	Rev AI	—	—
Gladia	ASR	—	—

Rates are editable USD per minute. Diarization may be bundled in some tiers; add your actual add-on rate if charged separately.

Cheapest option

At this volume and batch/realtime mix, the cheapest ASR option is — at roughly —/month.

Does not include storage, egress, model customization, or seat fees.

Audio-hour estimate

— hours of audio/month.

Useful for providers that price per hour or for capacity planning.

OpenAI Whisper API

Simple, widely supported batch transcription. No native diarization or streaming.

Best for: quick batch transcription

Deepgram Nova 2

Very fast batch and streaming. Diarization included on many tiers. Usage discounts at scale.

Best for: high-volume streaming or batch

AssemblyAI Universal

Strong diarization, PII redaction, and summarization as add-ons. Good for meeting assistants.

Best for: meetings with diarization

Google Cloud Speech-to-Text

Pay-per-15-seconds model. Enhanced model costs more; diarization included in some tiers.

Best for: GCP-native workloads

Azure Speech Services

Commitment tiers lower effective rate. Strong enterprise controls and custom model support.

Best for: Microsoft shops and custom models

Amazon Transcribe

Low per-minute rate; medical and call analytics are separate SKUs.

Best for: AWS-native call analytics

Rev AI

Premium human + ASR hybrid available. Higher per-minute cost for higher accuracy.

Best for: premium accuracy and human review

Gladia

Simple API with diarization and summarization bundled in some plans.

Best for: developer-first projects

How to use this estimate

Set your monthly audio minutes and the split between batch uploads and real-time streaming.
Add a diarization percentage if your use case needs speaker separation.
Edit per-minute rates to match your actual vendor quotes or enterprise tiers.
Compare the effective $/minute and total monthly cost columns to shortlist providers.

Cross-links: AI Voice Agent Cost Calculator, API Pricing Comparator, AI Startup Cost Calculator.