AccessDenied</Code> <Message> Invalid according to Policy:

[farski]

AccessDenied Invalid according to Policy:

from serverless-transcribe.

[farski]

@dariushazimi I believe Amazon Transcribe only supports audio files, so this uploader also only accepts audio files. A quick look at the docs make it seem like that's still the case. You'll have to strip the audio into a separate file before you upload.

[farski]

Yes I have. It looks like a good solution if you need a full featured analysis product. I only needed transcription, with a lightweight UI and that I could easily give access to others.

[dariushazimi]

@farski
Yes transcribe from AWS does support video files.

[farski]

@dariushazimi Do you have a list of formats that it supports?

[farski]

@dariushazimi The API only supports these formats: mp3, mp4, wav, flac based on the docs

[dariushazimi]

Correction. AWS Transcribe does support audio but I have tried the media solution analysis which supports video as well. Basically separates audio from video and does the transcription using the Transcribe. Works perfect but it has a lot of other services which I don't need and can be costly.

[dariushazimi]

Have you looked at the media analysis solution?

[dariushazimi]

Update: Uploaded an audio file successfully. But where is the transcription output ? :) I don't see any output on the page? checked the bucket and it is also empty?!!!

ps. Tried to DM you on twitter but I get: Sorry you can't message this person !!

[sachin10101998]

Can we process mp4 to mp3 conversion in lambda iteslf and allow people to upload mp4?

[farski]

@sachin10101998 Audio MP4 files should without requiring any conversion. If you're needing video MP4 files, those should also work but currently aren't allowed by the upload validation step. I can work on fixing that.

Actually sorry, I was confusing this with a different AWS service. Currently AWS Transcribe only supports audio files. Both AWS Transcribe and this app should allow for MP4 audio files, though. If you are having issues with that please open a new issue.