Hello William Oberst,
Your understanding is correct. As of today, Inuktitut is not supported for the Speech-to-Text service. Because this core transcription capability is not yet available, "speech-to-text translation" (transcribing Inuktitut audio and translating it to another language) is also not supported.
While the launch of Inuktitut Text-to-Speech signals Microsoft's active investment and partnership in Indigenous language preservation, no public release date has been announced for the Speech-to-Text feature.
Thanking Azar for prompt response and sharing the insight.
However, I will try to provide you with some recommendation steps you can try, here are the best paths forward for your research and for tracking this feature:
- Monitor Official Channels: This is the single source of truth for new language support. The STT table in this document will be updated the moment Inuktitut becomes available.
- Official Doc: Language and voice support for the Speech service
- Consider Immediate Workarounds for Your Research:
- Option A: Two-Step Pipeline (Manual + AI) The most reliable method today is to manually transcribe your recorded Inuktitut audio into text. You can then use the existing Azure AI Translator service to perform the text-to-text translation you require.
- Option B: Train a Custom Model If you have access to a significant amount of Inuktitut audio that also has accurate human-created transcriptions (e.g., 10+ hours), you may be able to train your own recognition model using the Azure AI Custom Speech service.
- Submit a Feature Request: This is the most effective way to communicate your specific research need directly to the engineering team. Strong use cases like yours help prioritize feature development.
- Feedback Portal: Azure AI + Machine Learning Feedback Forum
Please accept the answer and upvote for visibility to other community members.