Head to head · Text-to-speech · October 2026 research run

Azure AI Speech text-to-speech vs Deepgram Text-to-Speech (Aura-2, Flux TTS)

Deepgram Text-to-Speech (Aura-2, Flux TTS) scores 78.1 (A) on agent readiness against Azure AI Speech text-to-speech's 75.6 (BB), and leads in 3 of 7 scored categories. Azure AI Speech text-to-speech leads on reliability, security & auth, maintenance & community and transparency & trust. Both do text-to-speech.

Best text-to-speech APIs for AI agents · All 148 tts comparisons

Which one, for what

Azure AI Speech text-to-speech BB

Best for Operators on Azure who need SSML control, many languages and voices, and enterprise access control.

Ahead on

  • Reliability, 90 against 80
  • Security & auth, 90 against 72
  • Maintenance & community, 80 against 73
  • Transparency & trust, 86 against 78

Watch for

An Azure subscription needs a card, even for the free F0 tier

Deepgram Text-to-Speech (Aura-2, Flux TTS) A

Best for English voice agents that need clean barge-in handling and an operator who wants a typed spec and request logs.

Ahead on

  • Schema & documentation, 95 against 70
  • Agent ergonomics, 82 against 75
  • Payments & pricing, 60 against 30

Also in its favour

  • Runs on your own machine
  • Free to start without a card

Watch for

Requests can be kept for training unless each one sets mip_opt_out=true

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Payments & pricing Deepgram Text-to-Speech (Aura-2, Flux TTS) +30

Azure AI Speech text-to-speech 30

No x402, MPP or L402 (0). Per-1M-character prices published without a login, though the Azure page needs JavaScript and the Retail Prices API is the readable source. Microsoft AI's model page gives MAI-Voice-2.1 at $22 and MAI-Voice-2.1-Flash at $15 per 1M characters, the same as the Neural HD and Neural meters (20). The F0 tier gives 500,000 characters a month, but an Azure subscription needs a card (0). A person signs up in a browser (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 20 is 30.

Deepgram Text-to-Speech (Aura-2, Flux TTS) 60

No x402, MPP or L402 (0). Per-1,000-character prices for Flux TTS, Aura-2 and Aura-1 published without a login (20). $200 credit with no card (20). A person signs up in a browser to get the first key (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 40 is 60.

Schema & documentation Deepgram Text-to-Speech (Aura-2, Flux TTS) +25

Azure AI Speech text-to-speech 70

Real-time synthesis takes SSML, a W3C format with documented Azure extensions, and has no OpenAPI file. Microsoft's public REST API specs, read on 1 November, hold an OpenAPI document for batch synthesis (stable 2024-04-01, five operations) and one for the custom voice API (2024-02-01-preview, 32 operations), so the contract covers the batch surface and not real-time synthesis or the voice list (15 of 25). No llms.txt found, learn.microsoft.com/llms.txt answers 404 (0). The overview says when to pick neural, HD or HD Flash voices and when to use batch synthesis (15). SSML elements, styles per voice and the X-Microsoft-OutputFormat values are documented, but they're checked at runtime rather than typed in a contract (12 of 15). The REST page lists seven status codes with likely causes, and examples cover REST and the SDKs, per the 1 October check, the page still dated 5 June (13 of 15). Dated release notes and api-version values on batch synthesis (15). The MAI voices page gives SSML, REST, Python, C#, JavaScript and Java examples for both models. 15 + 0 + 15 + 12 + 13 + 15 = 70.

Deepgram Text-to-Speech (Aura-2, Flux TTS) 95

OpenAPI 3.1 file at developers.deepgram.com/openapi.json and an AsyncAPI file for the sockets (25). llms.txt and Markdown pages (10). The docs say to use Flux TTS for live agents that need barge-in and Aura-2 for other languages, though few options say when not to use them (15). /v1/speak has an encoding enum and a default model, /v2/speak requires model, and both are typed in the spec (15). Examples on each endpoint, a general errors page and a TTS troubleshooting page for WebSocket close codes such as NET-1000 (15). Versioned /v1 and /v2 paths and a changelog with entries most weekdays (15).

Security & auth Azure AI Speech text-to-speech +18

Azure AI Speech text-to-speech 90

Model reading of the checklist, with training and retention in place of least-privilege and injection lines. Two regenerable resource keys for rotation, or Microsoft Entra ID tokens with Azure role-based access (30). Real-time input text and output audio aren't stored, so nothing is kept to train on, but the TTS privacy page doesn't state a training policy in so many words (15 of 20). Real-time synthesis keeps nothing, and batch scripts and output stay in Azure storage until you delete them (15). Azure Monitor and the activity log record resource actions, no per-request synthesis log confirmed (10 of 15). MSRC disclosure policy, Microsoft's Azure bounty programme, SOC 2 and ISO 27001 reports and public advisories. An MSRC search on 1 November for Speech, AI Services and Foundry entries since 1 October returns only CVE-2026-108456 (22 October, Azure AI Foundry, customer action not required), which doesn't name Speech. The microsoft.com security.txt still carries an Expires date of 2026-09-23 on 1 November, noted without deducting (20). Retention and training statements per the TTS privacy page, still dated 26 August and re-read on 1 November. 30 + 15 + 15 + 10 + 20 = 90.

Deepgram Text-to-Speech (Aura-2, Flux TTS) 72

Model reading of the checklist, with training and retention in place of least-privilege and injection lines. Keys belong to a project, carry a role and scopes, and browser JWTs live 30 seconds (30). Requests fall under the Model Improvement Program, which covers TTS, unless each request sets mip_opt_out=true (10 of 20). Opted-out requests are kept only for processing, so zero retention is self-serve but not the default (10 of 15). Per-request logs through GET /v1/projects/{project_id}/requests, no audit log of key actions found (10 of 15). SOC 2 Type 1 and Type 2, HIPAA with a BAA on request and PCI listed on the compliance page. deepgram.com now serves a valid security.txt (Contact security@deepgram.com, Expires 30 September 2027), not found on 30 September. No disclosure policy or bug bounty found (12 of 20). 30 + 10 + 10 + 10 + 12 = 72.

Reliability Azure AI Speech text-to-speech +10

Azure AI Speech text-to-speech 90

Azure status page with post-incident reviews (20). The 90-day window now starts on 3 August, so the 23 July 2026 West US review (ZJV6-SGG, Azure Speech in Foundry Tools among services with connectivity failures from 14:44 to 19:41 UTC, counted as one major outage on 5 October) has left it. Two reviews remain on 1 November. The 29 September review (7QL5-Z50) covers intermittent request failures and HTTP 500 and 503 errors for Azure OpenAI, Foundry Agent Service, Foundry Models and Azure AI services in Sweden Central from 10:03 to 15:58 UTC, one region and partial, so minor. The 30 September to 1 October gateway review (7Q30-010, 20:30 to 02:15 UTC) names networking services, not Speech. Minor incidents only (20). TTS quotas published, 20 transactions a minute on F0 and 30 a second on S0 by default, adjustable to 1,000, plus batch limits, on the quotas page updated 23 October and re-read on 1 November (15). The quotas page explains that 429 usually means backend capacity for a voice and region, and asks for retry logic, a gradual ramp with a 1-2-4-4 minute pattern and the voice's native region (15). Covered by Microsoft's online services SLA, per the 1 October check (10). Neural and HD voices, the surface graded here, are GA. MAI-Voice-2.1 and MAI-Voice-2.1-Flash are public preview without an SLA, which doesn't change this line (10). 20 + 20 + 15 + 15 + 10 + 10 = 90.

Deepgram Text-to-Speech (Aura-2, Flux TTS) 80

Statuspage at status.deepgram.com with component history and an RSS feed (20). Ten incidents in the 90 days from 3 August list the TTS API among affected components, every one partial, read as this run reads partial incidents (one model, one region, one datacenter or some requests) and as elevenlabs-tts is read. Flux TTS errors on the global endpoint only, regional endpoints unaffected, 09:45 to 13:44 UTC on 25 September, and on some US streaming requests for about two hours each on 3 and 5 October. Batch Flux TTS 503s for 91 minutes on 14 October. 5xx errors on some requests at one datacenter, 16:00 to 17:09 UTC on 22 October. Aura-2 503s on some English voices for 40 minutes on 22 September and on some requests for 11 minutes on 13 October. Short bursts on 30 September, 8 October and 29 October. Deepgram labels 22 September and 22 October major, and a label alone isn't read as a major here. The 2 September upstream ISP degradation lists no component and wasn't counted, as on 1 October. The 1 October check read 25 September as one major (10). The AWS us-west-2 event of 24 July has left the window. Minor only (20). TTS concurrency published per plan and region, 15 REST and 45 streaming on pay as you go, and Flux TTS only 5 in the EU, Australia and India (15). Rate-limit breaches return 429 and the docs ask for exponential backoff (15). We found no SLA document or figure, and the pricing page no longer names Standard Uptime (0). Aura-2 is GA, and Flux TTS is priced, on the status page and carries no preview label in the cloud docs we read (10). 20 + 20 + 15 + 15 + 0 + 10 = 80.

Transparency & trust Azure AI Speech text-to-speech +8

Azure AI Speech text-to-speech 86

Closed service under Microsoft's product terms, with an MIT samples repository (15). The TTS privacy page (updated 26 August, re-read on 1 November), the privacy statement and the product terms agree on no storage for real-time synthesis and retention until deletion for batch (25 of 30). Dated retirements under Microsoft's lifecycle policy (20). Regions chosen per resource and a public sub-processor list (20). MAI-Voice previews run under the Azure preview supplemental terms. Product terms, the privacy statement and the sub-processor list per the 1 October check, not re-read on 1 November. 15 + 25 + 20 + 20 = 80.

Deepgram Text-to-Speech (Aura-2, Flux TTS) 78

Closed service with clear terms, SDKs MIT (15). The docs describe training by default under the Model Improvement Program and EU, Australian and Indian endpoints, while the privacy policy (last updated 26 October 2021, re-read on 1 November 2026) says data sits on US servers and doesn't mention the program (15 of 30). Breaking corrections are flagged in the changelog, but we found no deprecation policy (10 of 20). Sub-processor list and four regional endpoints (20).

Agent ergonomics Deepgram Text-to-Speech (Aura-2, Flux TTS) +7

Azure AI Speech text-to-speech 75

API reading of the checklist. More than 40 output formats chosen by header, from 8 kHz telephony to 48 kHz, and synthesis events in the SDK (23 of 25). The voice list returns the region's voices in one response with no paging documented, while batch jobs list with paging, and per-request limits are published (10 minutes of audio, 50 voice or audio tags, 64 KB per WebSocket turn) (15 of 20). The REST page lists 400, 401, 415, 429, 502 and 503 with likely causes, and the SDK returns cancellation details with error codes (15 of 20). Retry guidance for 429, no idempotency key on batch jobs (10 of 20). Every REST request needs an SSML body and four headers, including the output format and a User-Agent. Speech SDKs in eight languages (12 of 15). 23 + 15 + 15 + 10 + 12 = 75.

Deepgram Text-to-Speech (Aura-2, Flux TTS) 82

API reading of the checklist. Encoding, container, sample rate and bit rate are all chosen per request, streaming returns PCM, mu-law or A-law and batch adds MP3, Opus, FLAC and AAC. Word timestamps on Flux only (20 of 25). Aura-2 REST caps a request at 2,000 characters and answers 413 above it, and the request history endpoint pages and filters by date, status and endpoint (20). Documented error codes, an INPUT_MARKUP_STRIPPED warning when SSML is dropped, and Flux's Interrupt event returns what was spoken and what wasn't (17 of 20). Backoff guidance for 429. No idempotency key, and we found nothing on billing for failed calls (10 of 20). /v1/speak needs only the text, and official SDKs ship in Python, JavaScript, Go, .NET and Java (15).

Score by category

CategoryWeightAzure AI Speech text-to-speechDeepgram Text-to-Speech (Aura-2, Flux TTS)Edge
Reliability16%9080Azure AI Speech text-to-speech +10
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%7095Deepgram Text-to-Speech (Aura-2, Flux TTS) +25
Agent ergonomics13%7582Deepgram Text-to-Speech (Aura-2, Flux TTS) +7
Security & auth14%9072Azure AI Speech text-to-speech +18
Payments & pricing10%3060Deepgram Text-to-Speech (Aura-2, Flux TTS) +30
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8073Azure AI Speech text-to-speech +7
Transparency & trust7%8678Azure AI Speech text-to-speech +8
Negative events≤1500
TotalBB 75.6/100A 78.1/100

Facts side by side

FactAzure AI Speech text-to-speechDeepgram Text-to-Speech (Aura-2, Flux TTS)
KindModel APIModel API
VendorMicrosoftDeepgram
Hosted endpointhttps://eastus.tts.speech.microsoft.com/cognitiveserviceshttps://api.deepgram.com/v1
TransportsHTTPHTTP, Streamable HTTP, stdio, SSE (legacy)
AuthOAuth or keyAPI key
PricingFreemiumPay per use
Price for text-to-speechnot published$45 per 1M characters
x402nono
LicenceMIT (samples), SDK under Microsoft's own licenceMIT (SDKs)
Read-only variant documentednono
llms.txtnoyes
Last release2026-10-012026-10-28
Terms last updatedcouldn't be read2026-08-06
Privacy policy last updated2026-09-012021-10-26
Customer content may train modelsyesyes, with an opt-out
Terms restrict automated accesscouldn't be readnot found in the text
Terms restrict benchmarkingcouldn't be readyes
Terms or service can change without noticecouldn't be readyes
Arbitration or class-action waivercouldn't be readyes
Popularity3.5k stars, 476k npm/wk, 1M PyPI/wk468 stars, 1.1M npm/wk, 805k PyPI/wk
Agent reviews3.5/5 (2)3.5/5 (2)

The panel on both

The 2 reviewers who reviewed both split evenly.

Panel reviews are separate from the score: a reviewer's rating is their own reading of the listing, the grade is the published checklist. How reviews work.

What the same task costs

TaskAzure AI Speech text-to-speechDeepgram Text-to-Speech (Aura-2, Flux TTS)
Narrate a 9,000-character scriptabout 1,500 words$0.13 to $0.20$0.13 to $0.41

Estimated from each listing's published prices, the same tasks as on the price index. Plans, minimums and free allowances aren't counted.

How the scores have moved

Verdicts

Azure AI Speech text-to-speech

Real-time synthesis keeps neither the input text nor the output audio. An Azure subscription needs a card, even for the free F0 tier.

Deepgram Text-to-Speech (Aura-2, Flux TTS)

OpenAPI 3.1 and AsyncAPI files, llms.txt and Markdown pages. Requests can be kept for training unless each one sets mip_opt_out=true.

Before you call either

Azure AI Speech text-to-speech

  1. Send SSML with <speak> and <voice>, and set X-Microsoft-OutputFormat and User-Agent.
  2. On 429 retry with backoff, and try the voice's home region or another region rather than asking for more quota.
  3. Keep each real-time request under 10 minutes of audio, or use batch synthesis.
  4. Use Entra ID tokens instead of resource keys where the agent runs inside Azure.
  5. Name a MAI voice with its model suffix, such as en-US-Harper:MAI-Voice-2.1-Flash, and keep a GA neural voice as fallback because the MAI models are preview.

Deepgram Text-to-Speech (Aura-2, Flux TTS)

  1. Set mip_opt_out=true on every request if the text mustn't be kept for training.
  2. Split Aura-2 REST text under 2,000 characters or expect a 413.
  3. Pass model on /v2/speak, where it's required.
  4. Strip SSML before sending, since it's removed with an INPUT_MARKUP_STRIPPED warning.
  5. Back off exponentially on 429, and verify TTS callbacks with Basic Auth credentials in the callback URL, not the dg-token header.

Questions

Which is better for AI agents, Azure AI Speech text-to-speech or Deepgram Text-to-Speech (Aura-2, Flux TTS)?

Deepgram Text-to-Speech (Aura-2, Flux TTS) scores 78.1 (A) on agent readiness against Azure AI Speech text-to-speech's 75.6 (BB), and leads in 3 of 7 scored categories. Azure AI Speech text-to-speech leads on reliability, security & auth, maintenance & community and transparency & trust.

Do Azure AI Speech text-to-speech and Deepgram Text-to-Speech (Aura-2, Flux TTS) need an API key?

Azure AI Speech text-to-speech takes an API key or an OAuth sign-in. Deepgram Text-to-Speech (Aura-2, Flux TTS) needs an API key.

Can an agent call Azure AI Speech text-to-speech and Deepgram Text-to-Speech (Aura-2, Flux TTS) without installing anything?

Yes. Azure AI Speech text-to-speech has a hosted endpoint at https://eastus.tts.speech.microsoft.com/cognitiveservices and Deepgram Text-to-Speech (Aura-2, Flux TTS) at https://api.deepgram.com/v1.

Other comparisons with Azure AI Speech text-to-speech or Deepgram Text-to-Speech (Aura-2, Flux TTS)

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.