An Azure service that integrates speech processing into apps and services.
Khmer (km-KH) is documented for standard Azure text-to-speech, but the available Microsoft documentation does not explicitly confirm it for Personal Voice or Custom Neural Voice; submit the shared limited-access form before uploading recordings or planning production use.
Current support status
- Standard text-to-speech: Supported. Azure lists:
-
km-KH-SreymomNeural(female) -
km-KH-PisethNeural(male)
^1^
-
- Personal Voice: Not conclusively confirmed for Khmer by the published material available here. Personal Voice can synthesize more than 90 languages across more than 100 locales, but support must be checked against its feature-specific language list; general Khmer speech-to-text or standard TTS support does not establish Personal Voice support. Personal Voice uses automatic language detection rather than requiring a locale tag. ^2^
- Custom Neural Voice / professional voice fine-tuning: Also not conclusively confirmed for Khmer here. Custom Voice is a separate feature with its own supported-language, training-method, eligibility, and regional requirements. ^3^
Therefore, do not interpret the presence of Khmer standard voices as confirmation that Khmer recordings can be used to create a Personal Voice or fine-tuned Custom Neural Voice.
Official access route
There is no documented Khmer-specific registration program. Both features use the Custom Neural Voice and Custom Avatar Limited Access Review:
- Submit the official intake form: Limited Access Review.
- Describe the use case specifically as a Khmer cultural-tourism audio guide.
- State that:
- the speaker is the applicant;
- explicit consent is available;
- synthetic speech will be disclosed to listeners;
- a feedback/reporting channel will be provided.
- Ask Microsoft to confirm in writing:
- whether
km-KHis approved for Personal Voice; - whether it is approved for professional/Custom Neural Voice fine-tuning;
- which training and deployment region should be used.
- whether
Personal Voice—including its own-data demo and API—is limited access. Custom Neural Voice also requires registration, and the documented eligibility for Custom Neural Voice is limited to customers managed by Microsoft and working directly with a Microsoft account team. Approved usage is restricted to the use case accepted during registration. ^4^
Custom Neural Voice Lite is public preview, but access still follows approval, and broader deployment requires full Custom Neural Voice access. It is not documented as a Khmer preview program. ^4^
East Asia staging resource
East Asia currently lists:
- Personal Voice: supported
- Custom Voice deployment/use: supported
- Custom Voice training: not listed as supported
Consequently, the East Asia resource may be suitable for Personal Voice or an approved deployed custom voice, but it is not currently shown as a Custom Voice training region. A separate supported training region might be required if professional voice fine-tuning is approved. ^5^
Microsoft’s access documentation does not state that approval is permanently tied to retaining an existing Speech resource. Practical options are:
- Retain it temporarily if the intake form references that resource or if keeping it has acceptable cost and governance impact.
- Delete it if it is unused and recreate an appropriate resource after Microsoft confirms the feature, language, and training region. Resource-group deletion is irreversible, so delete only the Speech resource unless the whole group is disposable. ^6^
Continue withholding recordings until approval and Khmer eligibility are confirmed. For either custom-voice route, Microsoft requires a recorded consent statement and imposes biometric-data notice, consent, retention, and deletion responsibilities. ^4^
References
- Language and voice support for Azure Speech
- What is personal voice? - Foundry Tools | Microsoft Learn
- Custom voice overview - Speech service - Foundry Tools | Microsoft Learn
- Limited Access - Foundry Tools | Microsoft Learn
- Supported regions for Azure Speech
- Azure Resource Manager resource group and resource deletion