Last verified October 10, 2026.
Details about Microsoft Azure AI Speech come from its public pages as they were on October 10, 2026. Microsoft Azure AI Speech may have changed them since, and a page that was unclear to us may be clear to you. Corrections: support@readaloudai.org. We make ReadAloud, so read this page with that in mind and check anything important with Microsoft Azure AI Speech directly.
About Microsoft Azure AI Speech
How Microsoft Azure AI Speech positions itself: Microsoft's Speech service in Azure Foundry Tools, providing neural and HD text-to-speech voices in 100+ languages and locales with SSML, custom and personal voice options, and REST and WebSocket access.
Price per 1 million characters
USD per one million characters, by tier. Other units are shown as the vendor writes them, not converted.
| Option | Listed price | What it covers | Against ReadAloud Live |
|---|---|---|---|
| ReadAloud Live | $4 per 1M characters | the low-latency tier, built for live calls and voice agents | Reference point |
| ReadAloud Studio | $10 per 1M characters | the higher-priced tier for read-aloud and narration, with several English voices | $6 above ReadAloud Live |
| Microsoft Azure AI Speech: Neural | $15 per 1M characters (source: azure.microsoft.com) | Meter 'S1 Neural Text To Speech Characters'. Azure's page groups 'Neural/Neural HD Flash' on one price line; the mapping of this meter to that line is inferred. Pay-as-you-go rate, eastus region. | $11 higher than ReadAloud Live |
| Microsoft Azure AI Speech: Neural HD | $22 per 1M characters (source: azure.microsoft.com) | Meter 'Neural HD Text to Speech Characters', effective 2026-03-01. | $18 higher than ReadAloud Live |
| Microsoft Azure AI Speech: Personal Voice synthesis | $24 per 1M characters (source: azure.microsoft.com) | Voice creation listed as free on the pricing page; voice profile storage billed separately. | $20 higher than ReadAloud Live |
| Microsoft Azure AI Speech: Custom (Professional) voice, real-time | $24 per 1M characters (source: azure.microsoft.com) | Meter 'S1 Custom Neural Realtime Characters'; training and endpoint hosting billed separately (hosting $4.032 per model-hour per retail meter). Custom voice is a limited-access feature. | $20 higher than ReadAloud Live |
| Microsoft Azure AI Speech: Custom (Professional) voice, Neural HD | $48 per 1M characters (source: azure.microsoft.com) | Meter 'CNV Neural HD Synthesis Characters'. | $44 higher than ReadAloud Live |
| Microsoft Azure AI Speech: Long audio (batch) | $100 per 1M characters (source: azure.microsoft.com) | Meters 'S1 Neural Long Audio Characters' and custom equivalent. | $96 higher than ReadAloud Live |
| Microsoft Azure AI Speech: Commitment tiers (Neural) | Vendor's own unit: USD per month plus overage per 1M characters; not converted (source: azure.microsoft.com) | For example 80M tier $960 per month with overage $12.00 per 1M; 400M tier $3,900 per month with overage $9.75; 2000M tier $15,000 with overage $7.50; 4000M tier $24,000 with overage $6.00 (retail API, Azure-connected). | Not comparable: Microsoft Azure AI Speech lists a different unit |
| Microsoft Azure AI Speech: MAI-Voice | Vendor's own unit: USD per 1M characters (price not retrieved); not converted (source: azure.microsoft.com) | Listed on the pricing page; price not available in the retail API query. | Not comparable: Microsoft Azure AI Speech lists a different unit |
Every per-character list price we found for Microsoft Azure AI Speech, Neural at $15, Neural HD at $22, Personal Voice synthesis at $24, Custom (Professional) voice, real-time at $24, Custom (Professional) voice, Neural HD at $48 and Long audio (batch) at $100, is higher than both ReadAloud Live ($4) and ReadAloud Studio ($10) per 1M characters.
Microsoft Azure AI Speech's own headline, as of October 10, 2026: Neural voices $15 per 1M characters; Neural HD $22 per 1M characters (eastus retail list).
Free tier at Microsoft Azure AI Speech: Free (F0): Neural text to speech 0.5 million characters per month (pricing page); F0 rate limit 20 transactions per 60 seconds. ReadAloud: A one-time grant of free credits worth $0.10 (about 10,000 characters of speech), shared by all keys on the account.
Tiers cover different things, so compare on your own monthly volume, and check Microsoft Azure AI Speech's pricing page before you decide. Current ReadAloud prices: /developers.
Side by side
"Not stated on the pages we reviewed" marks what Microsoft Azure AI Speech's pages did not say. We print no head-to-head latency number.
| Topic | ReadAloud | Microsoft Azure AI Speech |
|---|---|---|
| Voices | Live: one American English voice. Studio: several English voices. | Not stated on the pages we reviewed (source: learn.microsoft.com) |
| Languages | English today | 100+ languages and locales for standard (neural) voices (source: learn.microsoft.com) |
| Custom voices or cloning | API, with a consent step and a payment method | Professional voice fine-tuning requires a recorded consent statement from the voice talent; custom voice is a limited-access feature per the pricing page. Personal voice is also offered ($24 per 1M characters synthesis; voice creation free). (source: learn.microsoft.com) |
| Streaming | WebSocket and HTTP streaming | WebSocket: yes; HTTP streaming: yes (source: learn.microsoft.com) |
| Time to first audio | ReadAloud Live: about 250 ms (our measurement, median to first audio byte, warm connection, San Jose; October 10, 2026; /developers#vs-elevenlabs) | Not stated on the pages we reviewed |
| Output formats | mp3, opus, wav, pcm; 8 kHz mu-law and A-law over WebSocket | mp3 (16, 24, 48 kHz), opus / ogg / webm (16, 24, 48 kHz), riff PCM wav (8 to 48 kHz), raw PCM (8 to 48 kHz), mulaw and alaw (8 kHz), amr-wb, g722, truesilk |
| SSML | No SSML and no audio tags. | Yes |
| OpenAI-compatible speech endpoint | Yes, /v1/audio/speech | No. REST text-to-speech takes SSML (Content-Type application/ssml+xml) posted to the resource's cognitiveservices endpoint with an X-Microsoft-OutputFormat header and subscription key or bearer token. Azure OpenAI neural voices are also selectable within the Speech service per the overview page, but no OpenAI-style speech route on the Speech service was read. |
| Limit per request | 5,000 characters | Audio length per request up to 10 minutes (longer is truncated); up to 50 distinct voice and audio tags in SSML; SSML message up to 64 KB per WebSocket turn (source: learn.microsoft.com) |
| Concurrency | Up to 12 Live streams per server; more start under load | Real-time TTS: F0 20 transactions per 60 seconds (not adjustable); S0 30 transactions per second default, adjustable up to 1,000 TPS. (source: learn.microsoft.com) |
| SDKs and plugins | Python and JavaScript libraries (readaloud), Pipecat and LiveKit plugins, any OpenAI SDK | Speech SDK, REST API and Speech CLI |
| HIPAA | Not claimed | Not stated on the pages we reviewed |
| SOC 2 | Not claimed | Not stated on the pages we reviewed |
| GDPR | Not claimed | Not stated on the pages we reviewed |
About Microsoft Azure AI Speech's compliance wording: No compliance certification statements were found on the speech pages read; the text-to-speech data, privacy and security page defines GDPR terms but is not a compliance claim. Microsoft publishes compliance material at platform level that was not reviewed.
What Microsoft Azure AI Speech lists as strengths
- Language coverage: standard neural voices in 100+ languages and locales. (source: learn.microsoft.com)
- Both REST and WebSocket synthesis, many output formats including 48 kHz and telephony mu-law and A-law. (source: learn.microsoft.com)
- Custom professional voice and personal voice options with consent requirements, plus Neural HD voices. (source: learn.microsoft.com)
- Commitment tiers list lower overage rates at volume (for example $6.00 per 1M on the 4000M tier). (source: prices.azure.com)
- Free F0 tier with 0.5 million Neural characters per month. (source: azure.microsoft.com)
Conditions to check with Microsoft Azure AI Speech
Limits or conditions from the pages we reviewed.
- Neural HD voices, personal voices and embedded voices do not support all SSML tags. (source: learn.microsoft.com)
- Custom voice is a limited-access feature. (source: azure.microsoft.com)
- Microsoft documents that most HTTP 429 errors with standard voices stem from limited backend capacity for a specific voice in the selected region rather than quota. (source: learn.microsoft.com)
Microsoft Azure AI Speech usage terms
From the vendor pages we reviewed; not legal advice.
- Attribution or conditions: Custom voices require voice-talent consent and are limited access. (source: learn.microsoft.com)
Where ReadAloud is different
- Voices: we could not read a voice count on Microsoft Azure AI Speech's pages, so we make no comparison of choice.
- SSML: Microsoft Azure AI Speech lists SSML support; ReadAloud does not accept it, so markup in your text would have to be removed or rewritten.
- Custom voices, as Microsoft Azure AI Speech's pages describe them: Professional voice fine-tuning requires a recorded consent statement from the voice talent; custom voice is a limited-access feature per the pricing page. Personal voice is also offered ($24 per 1M characters synthesis; voice creation free). (source: learn.microsoft.com)
Which one fits
Microsoft Azure AI Speech may suit you if this describes you: organizations on Azure or Microsoft 365 that need enterprise commitment tiers, many locales, and custom or personal voice options within the same cloud contract.
ReadAloud may suit you for streaming English speech at a low price per character.
What we could not confirm about Microsoft Azure AI Speech
- All Azure prices are single-source (Azure Retail Prices API, eastus); the pricing page does not show numbers in plain HTML and other regions may differ
- Whether the $15 'S1 Neural' meter is the same price line as 'Neural/Neural HD Flash' on the page
- MAI-Voice price
- Exact voice count
- Vendor-stated latency numbers (page describes how to measure first-byte latency but no figure was captured)
- SSML max size for REST calls
- HIPAA, SOC 2, GDPR statements for this product
- Commercial-use statement for generated audio
- OpenAI-compatible speech route on the Speech service (none found)
Sources
- Azure AI Speech pricing (renders prices client-side) (retrieved October 10, 2026)
- Azure Retail Prices API, Azure Speech meters, eastus, USD (retrieved October 10, 2026)
- Text to speech overview (retrieved October 10, 2026)
- Text to speech REST API reference (retrieved October 10, 2026)
- Speech service quotas and limits (retrieved October 10, 2026)
- Custom neural voice (retrieved October 10, 2026)
- Lower speech synthesis latency (retrieved October 10, 2026)