[In preview] Public Preview: MAI-Transcribe-1.5 in Microsoft Foundry model catalog
What the source says
Microsoft Foundry adds MAI-Transcribe-1.5 to the model catalog in public preview, Microsoft's next-generation speech-to-text model. The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency. Target workloads include call center analytics, media captioning, voice agent transcription, and meeting recording. MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models. It is available through Azure Speech. Learn more
Who is affected: Target workloads include call center analytics, media captioning, voice agent transcription, and meeting recording.
Why it matters: The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency.
Urgency: Monitor
Next action: Assess whether the customer's call center analytics, media captioning, voice agent transcription, or meeting recording workloads could benefit from the preview model.
Commercial opportunities
Discuss a preview evaluation through Azure Speech for workloads where accuracy on long-tail locales or transcription latency is important.
Technical actions
Validate Global Standard deployment availability, regional support, governance, and content safety requirements before proposing an evaluation.
Questions to ask the customer
Which of your transcription workloads are most constrained by word error rate or latency?
Do your workloads include long-tail locales where the previous generation's accuracy was insufficient?
Risks and objections
The model is in public preview, so production readiness and preview limitations should be validated before adoption.
Points to confirm
The effective date is unknown, so timing for customer follow-up is uncertain.
Reading for it_manager_dsi
Who is affected: Microsoft Foundry model catalog and Azure Speech
Why it matters: MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models.
Urgency: Monitor
Next action: Review whether current speech-to-text workloads align with the stated target workloads and evaluate a controlled preview test.
Commercial opportunities
Assess whether improved accuracy and faster transcription justify considering MAI-Transcribe-1.5 for relevant workloads.
Technical actions
Evaluate latency, word error rate, regional support, governance, and content safety controls in a representative test.
Questions to ask the customer
Which current workloads, if any, correspond to call center analytics, media captioning, voice agent transcription, or meeting recording?
Which regions, deployment controls, and latency requirements must be validated?
Risks and objections
Public preview status may limit production readiness and support expectations.
Points to confirm
The effective date is unknown, so timing for evaluation and adoption remains unconfirmed.
Reading for partner_channel
Who is affected: Target workloads include call center analytics, media captioning, voice agent transcription, and meeting recording.
Why it matters: MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models.
Urgency: Monitor
Next action: Review partner solutions involving call center analytics, media captioning, voice agent transcription, or meeting recording for potential fit with MAI-Transcribe-1.5.
Commercial opportunities
Assess whether partners can position MAI-Transcribe-1.5 for workloads requiring improved accuracy on long-tail locales or faster real-time and batch transcription.
Technical actions
Validate integration requirements and deployment coverage through Azure Speech and Global Standard deployments.
Questions to ask the customer
Which target workloads require speech-to-text transcription, and do long-tail locales or predictable latency matter?
Is public-preview availability acceptable for the intended workload?
Risks and objections
Public-preview status may require validation of readiness, support expectations, and production constraints before adoption.
Points to confirm
The effective date is unknown from the supplied source.
The supplied source does not establish production availability beyond public preview.
Reading for sales_manager
Who is affected: call center analytics, media captioning, voice agent transcription, and meeting recording
Why it matters: The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency.
Urgency: Monitor
Next action: Assess whether the listed target workloads align with current sales opportunities.
Commercial opportunities
Explore conversations around call center analytics, media captioning, voice agent transcription, and meeting recording.
Technical actions
Evaluate MAI-Transcribe-1.5 through Azure Speech for relevant workloads.
Review Global Standard deployments, regional support, governance, and content safety controls.
Questions to ask the customer
Which of your workloads involve call center analytics, media captioning, voice agent transcription, or meeting recording?
How important are accuracy on long-tail locales and predictable transcription latency to your requirements?
Risks and objections
Public preview status may limit readiness for customer commitments.
The effective date is unknown.
Points to confirm
The effective date is unknown from the supplied source.
Customer availability and production-readiness requirements have not been established.
Evidence and traceability
Each excerpt is linked to the primary source. Current raw version: 267.
Raw capture : 2026-08-21T06:15:09.102270+00:00 — hash 57205d9c3f036113…
Event
[In preview] Public Preview: MAI-Transcribe-1.5 in Microsoft Foundry model catalog Microsoft Foundry adds MAI-Transcribe-1.5 to the model catalog in public preview, Microsoft's next-generation speech-to-text model. The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency. Target workloads include call center analytics, media captioning, voice agent transcription, and meeting recording. MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models. It is available through Azure Speech. Learn more
[In preview] Public Preview: MAI-Transcribe-1.5 in Microsoft Foundry model catalog Microsoft Foundry adds MAI-Transcribe-1.5 to the model catalog in public preview, Microsoft's next-generation speech-to-text model. The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency. Target workloads include call center analytics, media captioning, voice agent transcription, and meeting recording. MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models. It is available through Azure Speech. Learn more
MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models.
[In preview] Public Preview: MAI-Transcribe-1.5 in Microsoft Foundry model catalog Microsoft Foundry adds MAI-Transcribe-1.5 to the model catalog in public preview, Microsoft's next-generation speech-to-text model.
The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency.
MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models.
[In preview] Public Preview: MAI-Transcribe-1.5 in Microsoft Foundry model catalog Microsoft Foundry adds MAI-Transcribe-1.5 to the model catalog in public preview, Microsoft's next-generation speech-to-text model. The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency. Target workloads include call center analytics, media captioning, voice agent transcription, and meeting recording. MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models. It is available through Azure Speech. Learn more
[In preview] Public Preview: MAI-Transcribe-1.5 in Microsoft Foundry model catalog Microsoft Foundry adds MAI-Transcribe-1.5 to the model catalog in public preview, Microsoft's next-generation speech-to-text model. The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency. Target workloads include call center analytics, media captioning, voice agent transcription, and meeting recording. MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models. It is available through Azure Speech. Learn more
[In preview] Public Preview: MAI-Transcribe-1.5 in Microsoft Foundry model catalog Microsoft Foundry adds MAI-Transcribe-1.5 to the model catalog in public preview, Microsoft's next-generation speech-to-text model. The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency. Target workloads include call center analytics, media captioning, voice agent transcription, and meeting recording. MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models. It is available through Azure Speech. Learn more
[In preview] Public Preview: MAI-Transcribe-1.5 in Microsoft Foundry model catalog Microsoft Foundry adds MAI-Transcribe-1.5 to the model catalog in public preview, Microsoft's next-generation speech-to-text model. The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency. Target workloads include call center analytics, media captioning, voice agent transcription, and meeting recording. MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models. It is available through Azure Speech. Learn more
The release adds improved accuracy with noticeably lower word error rate on long-tail locales than the previous generation, plus faster real-time and batch transcription with predictable latency.
MAI-Transcribe-1.5 is available via Global Standard deployments with Foundry-native governance, regional support, and content safety controls shared across MAI models.