fix: add audio_to_text and object_detection recipes to the catalog

[jeffmaury]

Fixes #695

What does this PR do?

Add 2 recipes to the catalog

Blocked by:

• Cannot integrate audio_to_text in AI Lab ai-lab-recipes#219
• object_detection recipe does not start in AI Lab ai-lab-recipes#220

Screenshot / video of UI

N/A

What issues does this PR fix or reference?

#695

How to test this PR?

Start the 2 new recipes:

• for the object_detection recipe, you can use this image: http://images.cocodataset.org/val2017/000000039769.jpg
• for the audio_to_text you can use the jfk.wav that is located in the data folder of the recipe repository

[vrothberg]

I do not see new recipes in the catalog. Am I missing something?

[lstocchi]

I do not see new recipes in the catalog. Am I missing something?

Be sure that you're actually using this PR. I'm running it now and i can see them. Maybe you forgot to build after switching branch?

[vrothberg]

I verified to be using the right branch and ran yarn install and yarn build before the yarn watch but do not see the recipe. Really strange.

[vrothberg]

aaaah ... I had to rm -rf ~/.local/share/containers/podman-desktop for it to show up.

[vrothberg]

Tested successfully 🏁 Great work!

[lstocchi]

Facing this issue on the audio to text. Which audio sample file did you use?
I downloaded a youtube video as WAV and used it

[axel7083]

aaaah ... I had to rm -rf ~/.local/share/containers/podman-desktop for it to show up.

I want to raise an important issue (again) justifying why you had to delete the folder, because on my computer too the new recipes was not visible

As I mentionned in #771 (comment) and available in #792

If the user at some point imported manually a model, which is the expected behaviour, he will not see new recipes after an update.

Because we have a unique catalog for the models and recipes, and we do not make any difference between user's one vs native. We ended up with a situation where the new recipes you added was not visible in my catalog because I have a copy of an old one.

[vrothberg]

I used the one from the local clone as Jeff mentioned in the PR description:

for the audio_to_text you can use the jfk.wav that is located in the data folder of the recipe repository

[vrothberg]

I want to raise an important issue (again).

@axel7083 that's what I must have hit as well.

[axel7083]

Images in markdown are not showing up

[lstocchi]

After reading the doc where it explains that we must use a 16-bit WAV file i was able to create a custom audio file and use it on the app. Maybe we should add this inside the app itself.
Like Upload an audio file you wish to have translated (it only supports a 16-bit WAV files)

BTW works fine. Just the problem with the images Axel mentioned.

[jeffmaury]

Facing this issue on the audio to text. Which audio sample file did you use? I downloaded a youtube video as WAV and used it

In the data folder of the recipe repository (see PR description)

[jeffmaury]

Images in markdown are not showing up

Will be addressed in a follow up PR when we will integrate recipes 1.0 as readme have changed on this side

[axel7083]

After reading the doc where it explains that we must use a 16-bit WAV file i was able to create a custom audio file and use it on the app. Maybe we should add this inside the app itself. Like Upload an audio file you wish to have translated (it only supports a 16-bit WAV files)

BTW works fine. Just the problem with the images Axel mentioned.

I opened containers/ai-lab-recipes#280 as this could be easy fixed by the application converting any audio to the expected format