AWS Transcribe Pricing 2026: $0.024/min Entry, $0.0078 at Scale

AWS Transcribe Standard costs $0.024 per minute at entry tier for both batch and streaming. That rate holds through your first 250,000 minutes per month, then drops across three further volume tiers. Depending on which tier you need (Standard, Medical, or Call Analytics), the final invoice can look very different. This post maps every tier, computes real monthly costs at several usage levels, and explains when the AWS ecosystem justifies the price, and when it does not.

This post is a linked deep-dive from our speech-to-text API pricing comparison.

AWS Transcribe Tier Pricing at a Glance

AWS Transcribe is three separate services with separate pricing clocks. Getting the tier wrong costs real money.

Standard Transcription Volume Tiers

Standard handles general-purpose audio in 100+ languages. Batch and streaming share the same rate:

Volume tiers apply to the combined total of batch and streaming minutes in a calendar month. If you process 200,000 batch minutes and 80,000 streaming minutes in January, all 280,000 minutes count toward your tier together, and the first 250,000 bill at $0.024 while the remaining 30,000 bill at $0.015.

Specialty Tier Rates

The Medical tier has the only meaningful batch-vs-streaming gap: batch is $0.075/min and streaming is $0.10/min. At scale, routing clinical audio to batch wherever latency allows saves 25%.

AssemblyAI U2

$0.0025/min

Deepgram Nova-3

$0.0077/min

AWS Standard

$0.024/min

AWS Medical

$0.075/min

AWS Call Analy.

$0.030/min

AWS Transcribe Standard vs leading alternatives at Tier 1/PAYG entry rate. Verified July 2026.

My take: the entry-tier gap between AWS Standard and Deepgram Nova-3 is nearly 3x. That does not automatically make AWS the wrong choice, but it does mean you need a concrete reason (S3 integration, IAM, compliance) to justify staying there at low-to-mid volume.

What the Tiers Mean in Practice

Standard Tier: Monthly Cost at Common Volumes

At Tier 1 ($0.024/min), the math is simple: multiply minutes by $0.024. The table below uses batch pricing, which equals streaming for the Standard tier:

The 5,000-hour and 10,000-hour rows cross into Tier 2. Flat $0.024 multiplication overstates them. At 5,000 hours, the first 4,167 hours bill at $0.024 and the remaining 833 hours bill at $0.015, producing $6,750, not $7,200.

Medical Tier: Real Costs for Healthcare Organizations

Medical is HIPAA-eligible and purpose-trained on clinical terminology, drug names, and anatomical shorthand. At $0.075/min for batch, it costs exactly 3.125 times the Standard rate.

A single physician dictating two hours of notes per day generates roughly 42 hours of audio per month: 42 hrs x 60 min x $0.075 = $189/month per physician. A clinic with 20 physicians runs about $3,780/month just in transcription fees before any infrastructure or integration costs.

For non-cardiology, non-primary-care specialties, test accuracy carefully. The general-purpose Standard tier with a well-tuned custom vocabulary sometimes produces acceptable accuracy at a third of the cost.

Call Analytics Tier: Post-Call vs Real-Time

Call Analytics includes turn-by-turn sentiment, issue detection, call categorization, and generative summarization. Tiered pricing matches Standard's structure but starts at $0.030/min:

The generative call summarization add-on costs an additional $0.0024/min at Tier 1. For contact centers already paying separately for NLP or summarization tools, the bundled pricing can be a genuine saving.

AWS Transcribe Free Tier

New AWS accounts receive 60 minutes of free transcription per month for the first 12 months. Standard, Medical, and Call Analytics each get their own 60-minute pool. Using all three in the same month gives you up to 180 free minutes total.

Key limits to know:

• Account-level, not user-level. If your organization has an existing AWS account, the 12-month clock started when that account was created.
• No hard cap. AWS bills automatically once you exceed 60 minutes. Set a CloudWatch billing alarm before you test anything.
• No rollover. Unused free minutes in January do not carry into February.
• Expiry is firm. After 12 months, billing begins from minute one.

Sixty minutes per month is enough to run integration tests and validate accuracy on your own audio. It is not enough for any production workload.

The AWS Ecosystem Advantage

The per-minute rate is not the whole cost story. Three factors can make AWS Transcribe cheaper than its headline rate suggests:

S3 integration eliminates egress fees. Batch transcription reads from S3 and writes results back to S3 at zero data transfer cost when both services are in the same region. Sending the same audio to an external API like Deepgram or AssemblyAI first requires downloading from S3 at $0.09/GB. One terabyte of audio per month adds $90 in transfer fees before the transcription bill. At scale, that gap narrows the price difference between AWS and cheaper competitors.

Lambda and EventBridge eliminate queue infrastructure. A Lambda function triggers transcription when audio lands in S3; EventBridge routes the completion event to downstream services. There is no queue to build or maintain. For high-volume pipelines with dozens of concurrent files, this saves meaningful engineering time.

IAM and CloudTrail handle compliance. Transcribe inherits AWS's identity model. Auditing all API calls, scoping permissions to specific roles, and satisfying SOC 2 or HIPAA requirements uses the same tooling your security team already manages. Building equivalent controls around an external API requires custom work: key rotation, request signing, network isolation, audit logging.

These advantages are real when all three conditions hold together. They are weak justification when you are on a different cloud, processing occasional files, or just starting out.

For a broader look at how AWS compares to other APIs, see our Deepgram vs AWS Transcribe comparison.

When AWS Transcribe Is Not the Best Fit

Price-sensitive teams at low-to-mid volume. At entry tier, AWS Standard at $0.024/min is about 3x Deepgram Nova-3 ($0.0077/min PAYG) and roughly 10x AssemblyAI Universal-2 ($0.0025/min). Volume tiers do help at scale, but competitors also have tiered and committed-use pricing that keep those gaps meaningful.

Teams not already on AWS. The ecosystem benefits above disappear entirely on GCP, Azure, or a mixed-cloud stack. You would be adding an AWS account, S3 buckets, IAM policies, and billing for the sole purpose of running transcription. That overhead is not worth it.

Workloads where per-second accuracy matters across many languages. AWS Transcribe is competitive but not the accuracy leader for every language or audio condition. Run benchmarks with your own audio before committing, especially for languages with limited training data.

Simple, low-volume use cases without API access. If you need to transcribe interviews or recordings without building API integrations, a tool like ConvertAudioToText handles file uploads, URL-based transcription, and export formats through a web interface with no AWS account or S3 bucket required. See the pricing page for flat-rate plans that remove per-minute billing entirely.

Hidden Costs to Watch For

S3 storage. Audio sitting in S3 costs $0.023/GB/month (S3 Standard). One thousand hours of 128 kbps MP3 is about 56 GB, costing roughly $1.29/month. Not alarming alone, but it accumulates if you archive at high volume.

Cross-region data transfer. S3 to Transcribe within the same region is free. Cross-region transfer is $0.02/GB. If your audio lives in us-east-1 and data residency rules push Transcribe to eu-west-1, that charge applies to every file.

CloudWatch logs. Transcribe outputs to CloudWatch, which charges $0.50/GB for log ingestion and $0.03/GB/month for storage. A high-volume pipeline with verbose logging can generate non-trivial CloudWatch costs. Set log retention policies before you scale.

Custom Language Model inference. Creating a custom language model is charged only when you use it on a job. This adds to the per-minute cost. Factor it into your total cost of ownership if CLMs are part of your accuracy strategy.

For more on total cost modeling, see hidden costs of transcription services and transcription pricing models explained.

Optimizing Your AWS Transcribe Spend

Route audio to the right tier. If your workload mixes clinical and general audio, implement routing logic. Sending non-clinical audio through Medical at $0.075/min instead of Standard at $0.024/min is a 3x cost penalty per minute.

Use batch where latency allows. Standard batch and streaming cost the same per minute, so latency tolerance is the only factor. But for Medical, batch at $0.075/min vs streaming at $0.10/min means routing to batch wherever feasible saves 25%.

Set CloudWatch billing alerts. Configure alarms at 50%, 80%, and 100% of your expected monthly spend. The free tier has no hard cap, and an errant integration test can exhaust it silently.

Compress audio before sending. Transcribe bills by duration, not file size. Sending MP3 at 128 kbps instead of uncompressed WAV costs nothing extra in transcription fees and cuts S3 storage and upload times. Pure savings.

Set S3 lifecycle rules. If you do not need to retain source audio after transcription completes, archive to S3 Glacier or delete on a schedule. Storage costs are linear and easy to forget.

FAQ

How much does AWS Transcribe cost per minute in 2026?

AWS Transcribe Standard costs $0.024 per minute for the first 250,000 minutes per month, dropping to $0.015/min up to 1 million minutes, $0.0102/min up to 5 million minutes, and $0.0078/min above that. Batch and streaming are priced identically at the Standard tier. Verify current rates on the official AWS Transcribe pricing page.

Does AWS Transcribe offer volume discounts?

Yes. Standard transcription uses four volume tiers: $0.024/min (0-250K min/mo), $0.015/min (250K-1M), $0.0102/min (1M-5M), and $0.0078/min above 5M. Volume tiers apply to combined batch and streaming usage. Call Analytics also has tiered pricing; the Medical tier does not publish volume discounts.

Is AWS Transcribe streaming more expensive than batch?

Not for the Standard tier. Both batch and streaming cost $0.024/min at Tier 1, and volume discounts apply to the combined total. The Medical tier does carry different batch ($0.075/min) and streaming ($0.10/min) rates, so for medical workloads streaming costs about 33% more than batch.

What is the AWS Transcribe free tier?

New AWS accounts receive 60 minutes of transcription per month for the first 12 months. This 60-minute allowance applies independently to Standard, Medical, and Call Analytics, so you can use all three without them competing for the same pool. After 12 months, or once you exceed 60 minutes in any month, billing starts automatically at the per-minute rate.

How does AWS Transcribe Medical pricing work?

AWS Transcribe Medical costs $0.075 per minute for batch and $0.10 per minute for streaming. That is 3.1 times the Standard batch rate. The Medical tier is HIPAA-eligible, supports primary care and cardiology specialties, and AWS will sign a Business Associate Agreement. Do not route clinical audio through the Standard tier if HIPAA compliance is required.

When should I choose a different STT provider over AWS Transcribe?

Consider alternatives when: you are not already on AWS (the ecosystem benefits do not apply), you need the lowest per-minute cost at low-to-mid volume (Deepgram Nova-3 at $0.0077/min is under a third of AWS's entry rate), or you want a no-API-key tool for occasional files. AWS Transcribe earns its keep when audio is already in S3, you need IAM-scoped access control, or Call Analytics features replace what you would otherwise pay for separately.

Sources

• Amazon Transcribe Pricing, official AWS pricing page, verified July 2026
• Deepgram Pricing, Nova-3 and Flux rates, verified July 2026
• AssemblyAI Pricing, Universal-2 and Universal-3.5 Pro rates, verified July 2026
• costbench.com, AWS Transcribe, tier thresholds cross-check
• costgoat.com, Amazon Transcribe, batch/streaming parity cross-check