Azure AI Speech text-to-speech (Neural, Neural HD, MAI voices)
Microsoft's enterprise speech synthesis service, with 200+ prebuilt neural voices across 80+ locales plus tiers for HD voices, multilingual voices, multi-talker voices and MAI-Voice models with emotion and style control.
- Weights
- Not released
- API
- Yes
- Commercial use
- Not established
Custom Neural Voice provides trained voice replication, and Voice Live covers realtime speech-to-speech. Speech-to-text covers 130+ locales.
The research notes for this model record other terms but do not state a commercial-use position, and no licence page was retrieved, so it is not established here. Check the vendor's own terms.
- Requires consent from the person depicted. Custom Neural Voice is a gated/limited-access capability under Microsoft's Responsible AI requirements, which historically require documented voice talent consent and an application process
Specification
- What it takes in
- text
- What it makes
- audio
- Controls
- emotion-control, style-prompting, voice-cloning, pronunciation-dictionary, speed-control, multi-speaker-dialogue, streaming-low-latency, timestamps
- Weights
- Not released
- API
- Yes
- Pricing
- not established
- Released or updated
- not established
- Languages
- 80, not individually listed by the vendor
Pricing is not established here. That means no page we could retrieve stated it, not that the model lacks one.
Strengths
- 200+ prebuilt neural voices across 80+ locales - the largest verified prebuilt roster here
- 60+ voices for en-US alone, including HD and multilingual turbo variants
- Multi-talker voices for conversational audio
- Speech-to-text in 130+ locales with fast transcription everywhere
- Enterprise compliance, regions and SLAs
Weaknesses
- Public pricing page renders all rates as '$-' placeholders; you must use the calculator or contact sales, so no price could be verified
- Voice tier naming (DragonHD, DragonHDOmni, MAI-Voice-1/2, multi-talker) is hard to navigate
- Custom Neural Voice is gated access, not self-serve
- Less expressive out of the box than ElevenLabs v3 or Hume Octave for character work
Best for
- enterprise localisation at scale
- regulated environments needing on-region processing
- broad language coverage
Where the sources disagree
Recorded rather than resolved. Picking the tidier number would hide the fact that the vendor's own pages do not agree.
- Pricing could not be verified: the Azure pricing page returns '$-' placeholders for Neural TTS, HD voices, Custom Neural Voice, speech-to-text and Voice Live. pricingModel is therefore null rather than estimated.
- VALL-E remains a Microsoft Research project with no productised API found on Azure pages; it is not offered as a purchasable model and is excluded from this dataset.
What to understand before using it
Explainers naming the mechanisms this model's own documentation mentions.
See it from the vendor
Shotcraft hosts no sample output. Judging a model means watching its own showreel with your own eyes, and a clip re-hosted here would be a clip we chose.
Nearby in the directory
Sonic 3.5
Cartesia's current TTS model, documented at sub-90ms first-byte latency across 42 languages. It is tuned for conversational delivery and for reading a
Voice and speech
XTTS-v2
An open-weights multilingual TTS and voice cloning model covering 17 languages at 24kHz, still widely used for local voice cloning. Coqui Inc. shut do
Voice and speech
Deepdub (eTTS, Speech-to-Speech, Live)
Enterprise AI voice and dubbing stack covering expressive text-to-speech with emotional control, speech-to-speech voice conversion and translation, vo
Voice and speech