Tasks and inputs
MAI-Voice-2.1 is listed for Text-to-speech. It accepts text, audio and produces audio. Use the publisher’s input format and preprocessing for your chosen deployment.
MAI-Voice-2.1 generates expressive multilingual speech from text through Azure Speech, with voice selection and supported custom-voice workflows.
Pricing conditions: The model page lists $22 per million characters. This is character billing, not a per-million-token rate. Azure account and regional terms apply.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
MAI-Voice-2.1 was announced on 1 October 2026. The standard revision is positioned for detailed expression and sustained narration. It supports multilingual speech generation and voice prompting through the MAI family’s supported access paths.
On Azure Speech, the model is selected through a voice name in SSML, including the MAI model suffix. A generic language-model request shape is not a substitute for the speech interface. Check the selected language and voice, emotion controls and any eligibility requirements for a custom voice on the account that will run the application.
The model page lists $22 per million characters. Character counts and language-model token counts are different units, so the figure cannot be compared directly with a text-token rate. Include revisions of the script and regenerated segments when estimating the cost of a finished recording. Listen to the final audio at ordinary playback speed before accepting it.
Use the same scripts across candidates, including uncommon names, acronyms, numbers and changes of tone. Check pronunciation, pauses and voice consistency throughout a longer recording. For interactive speech, measure the first playable audio and interruption behavior in the real application, rather than inferring latency from a model description.
Record the exact model identifier, deployment route and request settings with each result. Count complete runs, including rejected outputs, retries and tool calls. When the workflow creates files or code, open the files or execute the existing checks before accepting the result. A plausible first response can hide an incomplete final deliverable.
The specifications describe supported interfaces and request capacity. They do not establish a measured quality ranking or a latency guarantee for your account. SyncDev has not independently benchmarked this model. Choose a deployment after testing the languages, input formats and operating conditions that matter to your users.
Explore microsoft models, browse model tasks, or compare two models with the same acceptance criteria.
Compare the model on the work you need to complete.
MAI-Voice-2.1 is listed for Text-to-speech. It accepts text, audio and produces audio. Use the publisher’s input format and preprocessing for your chosen deployment.
Compare pronunciation, voice consistency and time to the first playable audio. Check the service’s billing unit for the actual script.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
Microsoft
MAI-Voice-2.1-Flash generates expressive multilingual speech from text through Azure Speech, with voice selection and supported custom-voice workflows.
Gemini 3.8 Flash-Lite TTS converts supplied text into speech through the Gemini API. It is a candidate for read-aloud features and repeated single-speaker audio jobs.
Gemini 3.8 Flash TTS turns a supplied text transcript into audio. It is a speech-generation model for narration and dialogue, with speaker and style information supplied separately from the words to be spoken.
xAI
Grok Voice TTS 1.0 from xAI: Convert text into spoken audio with a single API call. The API supports a rich set of expressive voices, inline speech tags for fine-grained delivery control, and output formats from high-fidelity MP3 to telephony-optimized μ-law.
OpenAI
GPT-Realtime-2.1 from OpenAI: Realtime speech-to-speech model with configurable reasoning, tool use, and robust voice-agent behavior.
Gemini 3.5 Live Translate Preview from Google: Low-latency audio-to-audio model for real-time speech translation across 70+ languages.