# Amazon Transcribe vs Deepgram

> Amazon Transcribe vs Deepgram: Deepgram is our pick, leading for developers and self-hosted. They split on free tier quota and published volume discounts.

![Amazon Transcribe logo](https://www.versusref.com/logos/amazon-transcribe.png) ![Deepgram logo](https://www.versusref.com/logos/deepgram.png)

Amazon Transcribe (AWS-native STT API with deep AWS ecosystem integration and compliance coverage) and Deepgram (Developer-first realtime STT API for voice agents and transcription at scale) are Speech-to-text APIs platforms, priced from $0.006/min and $0.0043/min respectively. Below: the bottom line, verified head-to-head facts, and real production costs.

Speech-to-text APIs platforms · 30 facts compared · all sourced · pricing verified Jul 20, 2026

## Bottom line

Deepgram wins three of four use cases and ties the fourth. For developers, it offers lower streaming prices at $0.0048 per audio minute versus $0.01 for Amazon Transcribe, higher default concurrency at 150 websocket streams versus 25, and a generous $200 credit free tier with no expiration or credit card required. For self-hosted deployments, Deepgram supports on-premises installation while Amazon Transcribe does not, a decisive advantage for teams with data-residency or latency requirements. For voice agents, Deepgram's lower streaming cost and self-host flexibility again tip the balance. Dictation ends in a tie.

## Verdict summary

| Use case | Winner | Why |
| --- | --- | --- |
| Developers | Deepgram | For developers building transcription products, batch pricing is the heaviest factor. |
| Dictation | Tie | The available facts cover API pricing, accuracy, and compliance for both Deepgram and Amazon Transcribe, but neither tool has verified information on the five attributes that matter most for a dictation use case: platform support for voice typing apps, local versus cloud audio processing, AI formatting and editing features, app pricing models, and app integrations for dictation workflows. |
| Self-Hosted | Deepgram | For a self-hosted deployment, the single most important question is whether the vendor allows on-premises installation at all. |
| Voice Agents | Deepgram | For voice agents, streaming latency is the top-weighted factor. |

## Amazon Transcribe vs Deepgram: head-to-head facts

| Fact | Amazon Transcribe | Deepgram | Verified |
| --- | --- | --- | --- |
| Batch price per audio minute | 0.006 $/audio-min ([source](https://aws.amazon.com/transcribe/pricing/)) | 0.004 $/audio-min ([source](https://deepgram.com/pricing)) | Jul 20 |
| Streaming price per audio minute | 0.01 $/audio-min ([source](https://aws.amazon.com/transcribe/pricing/)) | 0.005 $/audio-min ([source](https://deepgram.com/pricing)) | Jul 20 |
| Free tier quota | 60 min/mo for 12 months ([source](https://aws.amazon.com/transcribe/pricing/)) | $200 credit, no expiration, no credit card required ([source](https://deepgram.com/pricing)) | Jul 20 |
| WER (third-party benchmark) | 4.12 % WER ([source](https://artificialanalysis.ai/speech-to-text)) | 5.18 % WER ([source](https://artificialanalysis.ai/speech-to-text)) | Jul 20 |
| Streaming latency (vendor-claimed) | n/a | ~300 ms ([source](https://deepgram.com/product/speech-to-text)) | Jul 20 |
| Languages supported | 113 languages ([source](https://docs.aws.amazon.com/transcribe/latest/dg/supported-languages.html)) | ~50 ([source](https://deepgram.com/product/speech-to-text)) | Jul 20 |
| Speaker diarization | ✓  Included ([source](https://docs.aws.amazon.com/transcribe/latest/dg/diarization.html)) | ◑  Paid add-on ([source](https://deepgram.com/pricing)) | Jul 20 |

## Amazon Transcribe vs Deepgram pricing: true cost at 3 usage tiers

Monthly bill from published per-minute rates, batch and streaming separately. Sticker rates only; diarization and PII redaction add-ons price in the stack builder.

| Volume | Buyer | Amazon Transcribe | Deepgram | Note |
| --- | --- | --- | --- | --- |
| 1K min/mo (batch) | Side project | $6 | $4.30 | About 17 hours of audio. Free tiers may cover part of this. |
| 10K min/mo (batch) | Production app | $60 | $43 | About 167 hours of audio a month. |
| 100K min/mo (batch) | Call-center scale | $600 | $430 | About 1,667 hours. Most vendors negotiate volume rates here. |
| 1K min/mo (streaming) | Side project | $10 | $4.80 | About 17 hours of audio. Free tiers may cover part of this. |
| 10K min/mo (streaming) | Production app | $100 | $48 | About 167 hours of audio a month. |
| 100K min/mo (streaming) | Call-center scale | $1,000 | $480 | About 1,667 hours. Most vendors negotiate volume rates here. |

## Amazon Transcribe vs Deepgram: verdicts by use case

### Developers → Deepgram

For developers building transcription products, batch pricing is the heaviest factor. Deepgram charges $0.0043 per audio minute versus Amazon Transcribe at $0.006 per audio minute, a meaningful 28% cost advantage at scale. Streaming pricing follows the same pattern: Deepgram at $0.0048 per audio minute versus $0.01 for Amazon Transcribe, roughly half the cost. Both tools offer websocket streaming and word-level timestamps, so those attributes are neutral. On SDK coverage, Amazon Transcribe supports Python, JS, Java, .NET, Go, Ruby, PHP, C++, Rust, and CLI, while Deepgram covers JS/TS, Python, .NET, Go, Java, and Rust, giving Amazon Transcribe a slight edge in breadth. On audio formats, Deepgram supports 100+ formats versus Amazon Transcribe's eight listed batch formats, giving Deepgram the edge there. The pricing advantage on the top-weighted attribute, combined with broader format support, tips the decision.

### Dictation → Tie

The available facts cover API pricing, accuracy, and compliance for both Deepgram and Amazon Transcribe, but neither tool has verified information on the five attributes that matter most for a dictation use case: platform support for voice typing apps, local versus cloud audio processing, AI formatting and editing features, app pricing models, and app integrations for dictation workflows. Deepgram does offer a self-host option while Amazon Transcribe does not, which is a slight advantage for local processing, but this is an API-level detail rather than a consumer dictation app feature. Without facts on dictation platforms, formatting features, or app-level pricing, a meaningful winner cannot be determined.

### Self-Hosted → Deepgram

For a self-hosted deployment, the single most important question is whether the vendor allows on-premises installation at all. Deepgram explicitly supports a self-host and on-prem option, while Amazon Transcribe does not offer any self-host option. This difference alone is decisive at the heaviest attribute weight. All other attributes in this use case, including hardware requirements, model weights licensing, and maintenance status, are secondary to this fundamental gate. Because Amazon Transcribe is cloud-only, a buyer who needs to run the model on their own hardware cannot use it regardless of pricing or accuracy.

### Voice Agents → Deepgram

For voice agents, streaming latency is the top-weighted factor. Deepgram publishes a 300 ms streaming latency figure, while Amazon Transcribe has no published latency claim to compare against. Both offer WebSocket streaming APIs, so that attribute is tied. On streaming price, Deepgram charges $0.0048 per audio minute versus Amazon Transcribe at $0.01 per audio minute, making Deepgram roughly half the cost for live streams. On concurrency, Deepgram offers 150 WebSocket concurrent connections on the base pay-as-you-go plan, compared to 25 concurrent streams for Amazon Transcribe by default. Both support custom vocabulary. Across the three decisive differentiators, streaming latency, streaming price, and concurrency, Deepgram leads on all three.

## Amazon Transcribe vs Deepgram: common questions

### Is Deepgram cheaper than Amazon Transcribe?

For batch transcription, Deepgram charges about $0.0043 per audio minute versus Amazon Transcribe's $0.006, making Deepgram notably cheaper. For streaming, Deepgram is again lower at about $0.0048 per audio minute compared to Amazon Transcribe's $0.01. One exception worth noting: speaker diarization is included with Amazon Transcribe, while Deepgram charges an extra $0.002 per audio minute for it. So the better value depends on which features you actually use.

### Does Deepgram or Amazon Transcribe cost less for streaming transcription?

Deepgram charges about $0.0048 per audio minute for streaming, while Amazon Transcribe charges $0.01 per audio minute. At scale, Deepgram's streaming rate is meaningfully lower. Both services use usage-based pricing, so your bill scales directly with audio volume.

### Is Deepgram or Amazon Transcribe better for multilingual applications?

Amazon Transcribe supports 113 languages, significantly more than Deepgram's 50 languages. Both tools offer automatic language detection. If broad language coverage is a priority, Amazon Transcribe has a clear advantage.

### Is Deepgram or Amazon Transcribe better for HIPAA-compliant healthcare applications?

Both Deepgram and Amazon Transcribe offer a HIPAA BAA, SOC 2 Type II certification, and GDPR or EU data residency support. Either can meet standard healthcare compliance requirements, so the decision can hinge on other factors like pricing or self-hosting needs.

### Which service has a better free trial for testing speech-to-text?

Amazon Transcribe offers 60 free minutes per month for the first 12 months. Deepgram provides a $200 credit with no expiration date and no credit card required to start. This makes Deepgram's trial more flexible for extended evaluation.

### Can I self-host Amazon Transcribe or Deepgram on my own infrastructure?

Deepgram supports self-hosting and on-premises deployment, while Amazon Transcribe does not offer a self-host option. If keeping audio data within your own infrastructure is a hard requirement, Deepgram is the only viable choice between the two.

### Does Amazon Transcribe include speaker diarization, or is it an extra cost?

Amazon Transcribe includes speaker diarization at no additional charge beyond the base transcription rate. Deepgram, by contrast, treats diarization as a paid add-on, costing about $0.002 per audio minute extra.

### How does accuracy compare between Deepgram and Amazon Transcribe based on third-party benchmarks?

In third-party benchmarks, Amazon Transcribe achieved a 4.12% word error rate while Deepgram scored 5.18%. Because a lower word error rate indicates higher accuracy, Amazon Transcribe holds a slight edge in these independent tests.

### Does Deepgram support entity detection, and how does it compare to Amazon Transcribe on this feature?

Deepgram supports entity detection, while Amazon Transcribe does not. If your workflow requires identifying named entities directly from transcripts, Deepgram has a clear advantage.

Source: https://www.versusref.com/stt/amazon-transcribe-vs-deepgram/