Hi Team,
Following are the queries that We would like to know about the NVIDIA Conversational AI:
- Does your service provide Hindi text to English text conversion or Hindi conversation to English Text?
- Is there an option for speaker Diarization for the Hindi model?
- Also, is there an option to detect the silence and its duration in the conversation?
- Can we know the Tone of the speaker for each sentence? For example: Is the speaker’s tone high or low?
Kindly, go through the points and provide us with the service details if available.
Thank you.
Hi @iamgarimanarang
Thanks for your interest in Riva,
- Regarding Hindi we have ASR (Speech recognition), But currently don’t have TTS (will check with my team),
Please find the below link for reference
The Making of RIVA Hindi ASR Service — NVIDIA Riva
Regarding the other question I will check with my team and get back
Thanks
1 Like
Hi @iamgarimanarang
Sincere Apologies for the long delay,
I have answers for the team
Q: Does your service provide Hindi text to English text conversion or Hindi conversation to English Text?
A: Hindi text → English text and Hindi Speech → English text are not available publicly today. Would you be interested in early access software?
Q: Is there an option for speaker Diarization for the Hindi model?
A: Speaker Diarization is supported only for English today
Q: Also, is there an option to detect the silence and its duration in the conversation?
A: You can set enable_word_time_offsets (gRPC & Protocol Buffers — NVIDIA Riva) to true to get the start and end time offsets (timestamps) for words in the transcript
Q: Can we know the Tone of the speaker for each sentence? For example: Is the speaker’s tone high or low?
A: This is not supported today