Use Cases
FUTO Voice Input currently outputs American spelling and is tuned to American pronunciation. For users with device language set to English (Australia) or English (GB), the transcription consistently produces American spellings (e.g., color, organize, traveling, license) instead of Commonwealth spelling.
This creates usability issues in everyday writing, work communication and any context where correct Australian/British spelling is required. It also reduces recognition accuracy for these accents, since the model is not trained on Australian or British phonetics.
The core problem is that there is no available English model that matches Commonwealth spelling or Australian/British pronunciation, even when the system language is set accordingly.
Proposal
Add support for Australian English (en‑AU) and/or UK English (en‑GB) speech‑to‑text models. Either variant would resolve the spelling issue, as Australian and British spelling conventions are effectively identical for most words. Both would also improve recognition accuracy for users with Australian, British, or mixed accents.
How this could look in the application
- In FUTO Voice Input → Languages / Models, provide downloadable models for:
- English (Australia)
- English (United Kingdom)
- When the device language is set to en‑AU or en‑GB, FUTO Voice Input should automatically select the matching model.
- Users could still manually choose a model if they prefer a different variant.
Notes on precedent
Many widely used speech engines (including Google, Apple, and Microsoft) offer separate en‑AU and en‑GB models. These variants provide correct Commonwealth spelling and improved recognition for local accents. Adding similar options would bring FUTO Voice Input in line with common behaviour across other speech‑to‑text systems.
I’m happy to provide sample audio or help test early builds if that would be useful.
References
No response
Use Cases
FUTO Voice Input currently outputs American spelling and is tuned to American pronunciation. For users with device language set to English (Australia) or English (GB), the transcription consistently produces American spellings (e.g., color, organize, traveling, license) instead of Commonwealth spelling.
This creates usability issues in everyday writing, work communication and any context where correct Australian/British spelling is required. It also reduces recognition accuracy for these accents, since the model is not trained on Australian or British phonetics.
The core problem is that there is no available English model that matches Commonwealth spelling or Australian/British pronunciation, even when the system language is set accordingly.
Proposal
Add support for Australian English (en‑AU) and/or UK English (en‑GB) speech‑to‑text models. Either variant would resolve the spelling issue, as Australian and British spelling conventions are effectively identical for most words. Both would also improve recognition accuracy for users with Australian, British, or mixed accents.
How this could look in the application
Notes on precedent
Many widely used speech engines (including Google, Apple, and Microsoft) offer separate en‑AU and en‑GB models. These variants provide correct Commonwealth spelling and improved recognition for local accents. Adding similar options would bring FUTO Voice Input in line with common behaviour across other speech‑to‑text systems.
I’m happy to provide sample audio or help test early builds if that would be useful.
References
No response