Add speech to text sample #24
Conversation
Signed-off-by: Radu Matei <[email protected]>
|
is there a reason we want to check in the |
Signed-off-by: Radu Matei <[email protected]>
|
Removed and added it to gitignore. |
|
Is this example ready to merge? |
| The transcription service is a service running the OpenAI Whisper model on an Nvidia RTX 4000 Ada Generation GPU on | ||
| an LKE cluster. |
There was a problem hiding this comment.
| The transcription service is a service running the OpenAI Whisper model on an Nvidia RTX 4000 Ada Generation GPU on | |
| an LKE cluster. | |
| The transcription service is a service running the OpenAI Whisper model on an | |
| Nvidia RTX 4000 Ada Generation GPU on an LKE cluster. |
spacing nit
| [variables] | ||
| # This is a currently deployed transcription API endpoint | ||
| # This may not be available by the time you run this sample. | ||
| transcription_api = { default = "http://172.236.120.221:30001/v1/audio/transcriptions" } |
There was a problem hiding this comment.
As warned in the comment above, I don't think this service is available as of testing today. The api/transcribe internal endpoint returns a 504, seen when inspecting the browser console:
/api/transcribe:1
Failed to load resource: the server responded with a status of 504 (Gateway Time-out)
Interestingly, the app logs via spin aka logs don't show any explicit errors, despite captions/src/index.ts containing error handling, so that might be another item to revisit.
There was a problem hiding this comment.
If we do plan on standing the inferencing service up again and merge this example, we might consider adding minimal 'how to build and deploy' steps here.
No description provided.