# Amazon Transcribe

AWS speech-to-text service that transcribes audio files from S3 or live streams into JSON transcripts with timestamps.

- Provider: Amazon Web Services (AWS) (https://aws.amazon.com/transcribe/)
- Category: Transcribe audio
- Type: Hosted API
- Pricing model: Pay as you go, billed monthly per second of audio [3]
- Free tier: 60 audio minutes a month for the first 12 months [1]
- Authentication: AWS IAM credentials with SigV4-signed requests [10]
- Connect via: REST API through AWS SDKs and CLI, HTTP/2 and WebSocket streaming, general AWS MCP server [6]
- Read from the provider's pages on 2026-11-04
- Page: https://supermarket.dev/tools/transcribe-audio/amazon-transcribe

## Pricing

### Billing
- Pricing model: Pay as you go, billed monthly per second of transcribed audio [3]
- Free tier: 60 audio minutes a month for the first 12 months; not applied to content redaction or custom language models [1]
- Card required for free tier: Not published
- Minimum spend: Not published
- Billing period: Monthly; usage charged in 1-second increments with no minimum duration [3]

### Prices
US East (N. Virginia) rates from the AWS price list. Per hour = per-second price x 3,600: batch $0.0001 x 3,600 = $0.36; streaming $0.0001667 x 3,600 = $0.60. The pricing page lists the same $0.006 and $0.01 per minute. No volume tiers for these two.
| Mode | $ per hour of audio | $ per minute | $ per second |
|---|---|---|---|
| Standard batch (S3 files) | $0.36 | $0.006 | $0.0001 |
| Standard streaming | $0.60 | $0.01 | $0.0001667 |
Source [2]

### Other charges
US East (N. Virginia), first tier (first 250,000 minutes a month); lower tiers cost less. Hourly = per-second x 3,600.
| Item | $ per hour of audio | $ per minute | $ per second |
|---|---|---|---|
| Content redaction, batch and streaming (added to standard) | $0.144 | $0.0024 | $0.00004 |
| Toxicity detection (added) | $0.216 | $0.0036 | $0.00006 |
| Custom language model, batch (added) | $0.36 | $0.006 | $0.0001 |
| Call Analytics, post-call and real-time | $1.80 | $0.03 | $0.0005 |
| Medical, batch and streaming | $4.50 | $0.075 | $0.00125 |
Source [2]

## Tools

### MCP server
The awslabs/mcp repository holds no Transcribe-specific server; Transcribe is reached through the general AWS MCP server.
- Official MCP server: Yes, remote [15]
- Address or package: `https://aws-mcp.us-east-1.api.aws/mcp` (via `mcp-proxy-for-aws-cli`) [15]
- Transcribe-specific MCP server: No [17]

### MCP tools
General AWS MCP server, not Transcribe-specific. tools/list returned 8 tools without credentials. Names carry the prefix `aws___`.
| Tool | What it does | Read-only |
|---|---|---|
| `aws___run_script` | Runs Python in a sandbox with AWS API access through `call_boto3` | No |
| `aws___get_presigned_url` | Creates pre-signed S3 URLs for uploads and downloads | No |
| `aws___get_tasks` | Polls the status of long-running tasks | Yes |
| `aws___search_documentation` | Searches AWS documentation | Yes |
| `aws___read_documentation` | Fetches full AWS doc pages as markdown | Yes |
| `aws___retrieve_skill` | Retrieves an AWS skill (workflows, references) | Yes |
| `aws___list_regions` | Lists all AWS regions | Yes |
| `aws___get_regional_availability` | Shows AWS resource availability per region | Yes |
Source [17]

### API
| Endpoint | What it does |
|---|---|
| `StartTranscriptionJob` | Starts a batch job on a media file in S3 |
| `GetTranscriptionJob` | Returns the status of a job and the transcript location |
| `ListTranscriptionJobs` | Lists jobs matching criteria |
| `DeleteTranscriptionJob` | Deletes a job |
| `StartStreamTranscription` | Starts a bidirectional HTTP/2 or WebSocket stream for live audio |
| `CreateVocabulary` | Creates a custom vocabulary |
Source [13]

### Features
- Default model: Not published
- Languages: Over 100 language and dialect codes listed (see table); support differs for batch and streaming [6]
- Speaker labels: Yes, batch and streaming; up to 30 speakers (`spk_0` to `spk_29`) [5]
- Word timestamps: Yes; start time, end time and confidence score for every word [4]
- Live streaming: Yes, over HTTP/2 and WebSocket [9]
- Max file size: 2 GB or 8 hours per batch call [9]
- Audio input: Batch: AMR, FLAC, M4A, MP3, MP4, Ogg, WebM, WAV from S3. Streaming: FLAC, Ogg Opus, PCM. One or two channels [4]

## Authentication

### Methods
| Method | Used for | How |
|---|---|---|
| AWS IAM credentials with Signature Version 4 | All API calls, batch and streaming | SDK or CLI signs requests with IAM user keys or temporary role credentials |
| S3 permissions | Batch jobs reading audio and writing transcripts | Media is read from S3; the output bucket must grant Transcribe write permission |
| OAuth or API key | Not offered by Transcribe | Not published |
Source [10]

### Getting access
- Who creates the account: A person with an AWS account; sign-up by an agent alone is not stated [10]
- Where to get a key: No Transcribe-specific key; AWS IAM credentials (temporary role credentials preferred over long-term user keys) [10]
- Other requirements: An AWS account and an Amazon S3 bucket holding the media for batch jobs [12]

## Setup

### Addresses
- REST API: `transcribe.us-east-1.amazonaws.com` (batch), `transcribestreaming.us-east-1.amazonaws.com` (streaming); other regions use the same pattern [8]
- MCP server: `https://aws-mcp.us-east-1.api.aws/mcp` (general AWS MCP server) [15]
- OpenAPI spec: Not published
- Docs for agents: `https://docs.aws.amazon.com/transcribe/latest/dg/llms.txt`; API reference `https://docs.aws.amazon.com/transcribe/latest/APIReference/llms.txt` [14]
- Status page: `https://health.aws.amazon.com/health/status` (HTTP 200) [17]

### Install
| What | Command |
|---|---|
| Start a batch job with the AWS CLI | `aws transcribe start-transcription-job --region us-west-2 --transcription-job-name my-job --media MediaFileUri=s3://BUCKET/file.flac --language-code en-US` |
| Connect to the AWS MCP server (Kiro config) | `uvx mcp-proxy-for-aws-cli@latest https://aws-mcp.us-east-1.api.aws/mcp` |
| AWS CLI setup for agents | `aws configure agent-toolkit` |
| Claude Code plugin | `/plugin install aws-core@claude-plugins-official` |
| AWS SDKs | Listed for .NET, C++, Go, Java V2, JavaScript, PHP V3, Python Boto3, Ruby V3, Rust; no install command on the docs page |
Source [12]

## Limits

### Rate limits
| Operation | Default per second per region | Adjustable |
|---|---|---|
| `StartTranscriptionJob` | 25 | Yes |
| `GetTranscriptionJob` | 30 | Yes |
| `ListTranscriptionJobs` | 5 | Yes |
| `DeleteTranscriptionJob` | 5 | Yes |
| `StartStreamTranscription` | 25 | Yes |
| `StartStreamTranscriptionWebSocket` | 25 | Yes |
Source [8]

### Other limits
- Concurrent batch jobs: 250 per region (adjustable) [8]
- Concurrent streams: 25 per region (adjustable) [8]
- Maximum audio file length: 28,800 seconds (8 hours); Call Analytics batch 14,400 seconds [8]
- Maximum audio file size: 2 GB; Call Analytics batch 500 MB [8]
- Minimum audio file duration: 500 milliseconds [8]
- Audio channels: 2 [8]
- Streaming connection length: Up to 8 hours [9]

## Privacy

### Your data
- Data location: Batch transcripts go to your S3 bucket or a service-managed bucket; processing region is the one you call [4]
- Retention: Transcripts in a service-managed bucket and job records are kept 90 days; Amazon may temporarily store content [4]
- Used for training: Content may be used to improve the service; opt-out through an AWS Organizations policy [9]
- Zero data retention: Not offered as a setting; deletion of stored content is requested through a Support case [4]

### Compliance
- Certifications: Not published
- HIPAA: Eligible under the AWS HIPAA BAA; customers must encrypt PHI at rest and in transit [3]

## Sources

1. Amazon Transcribe pricing: https://aws.amazon.com/transcribe/pricing/
2. AWS price list, Amazon Transcribe, us-east-1: https://pricing.us-east-1.amazonaws.com/offers/v1.0/aws/transcribe/current/us-east-1/index.json
3. What is Amazon Transcribe?: https://docs.aws.amazon.com/transcribe/latest/dg/what-is.html
4. Data input and output: https://docs.aws.amazon.com/transcribe/latest/dg/how-input.html
5. Partitioning speakers (diarization): https://docs.aws.amazon.com/transcribe/latest/dg/diarization.html
6. Supported languages and language-specific features: https://docs.aws.amazon.com/transcribe/latest/dg/supported-languages.html
7. Amazon Transcribe features: https://docs.aws.amazon.com/transcribe/latest/dg/feature-matrix.html
8. Amazon Transcribe endpoints and quotas: https://docs.aws.amazon.com/general/latest/gr/transcribe.html
9. Amazon Transcribe FAQs: https://aws.amazon.com/transcribe/faqs/
10. Identity and Access Management for Amazon Transcribe: https://docs.aws.amazon.com/transcribe/latest/dg/security-iam.html
11. Data protection in Amazon Transcribe: https://docs.aws.amazon.com/transcribe/latest/dg/data-protection.html
12. Transcribing with the AWS CLI: https://docs.aws.amazon.com/transcribe/latest/dg/getting-started-cli.html
13. Amazon Transcribe API Reference (llms.txt): https://docs.aws.amazon.com/transcribe/latest/APIReference/llms.txt
14. Amazon Transcribe Developer Guide (llms.txt): https://docs.aws.amazon.com/transcribe/latest/dg/llms.txt
15. Agent Toolkit for AWS (GitHub README): https://github.com/aws/agent-toolkit-for-aws
16. AWS services in scope by compliance program: https://aws.amazon.com/compliance/services-in-scope/
17. Checked by supermarket.dev