IssueTracker

Please describe your requested enhancement. Good feature requests will solve common problems or enable new use cases.
What you would like to accomplish:
We would like to have Devanagari notation for hi-IN fully supported for the model Chirp_2, as seen in the attached image, it does a mix between Roman characters and Devanagari script.

How this might work:
It seems it might require an additional normalizer to convert from Latin (produced by the model) to Devanagari.

If applicable, reasons why alternative solutions are not sufficient:
The alternatives are using other models that do support Devanagari notation for hi-IN, however from testing it seems that the model Chirp_2 is more accurate for picking up words from difficult audios. Also Chirp_2 has more updated features.

Other information (workarounds you have tried, documentation consulted, etc):
As a workaround, change the model to either “Long” or “Chirp” (Chirp 1),  and they have written the text in Hindi as intended with the Devanagari script [1]. But it is not as accurate.

[1] https://cloud.google.com/speech-to-text/v2/docs/transcription-model (Available models)
[2] https://cloud.google.com/speech-to-text/v2/docs/best-practices (Best practices)