Custom ASR

Contact us at [e-mail] to enable this feature in your workspace.

Example API Code

To test this code, you can create a live endpoint using codesandbox.io. Click here to open an example from us.

Once you make and save any modifications to that CodeSandbox code, a fork of the codebase will be automatically created for you to modify and test around further.

Configuring the Sample Custom API on Project Creation Wizard

First, you will need to get the base URL of the API from the CodeSandbox UI. It should be available on the URL bar on the preview panel on the right side.

You may need to expand the panel first if it is collapsed by default.

In the sample API that we provided, there are two HTTP POST endpoints.

1. {baseUrl}/text-extraction/example-text, which returns the transcription as plain text

2. {baseUrl}/text-extraction/example-json, which returns the transcription as Importable JSON

Once you have the URL, you can create, store, and use Custom API when creating a new project.

1. In the “Active projects” page, click on “Create project”

2. In Step 1, “Upload”, upload your image/pdf for OCR, or audio files for ASR, then click “Next”

3. Once you’re on Step 2, “Preview”, you can click on the dropdown labeled “Apply OCR Method” (for image/pdf files) or “Apply ASR Method” (for audio files).

4. After you click to open it, you should see a button to “+ Add new API…”

5. A dialog will appear with 3 fields.

   1. Fill the “Name” field with the desired name for your Custom API.

   2. Fill the “Custom API URL” with one of the two endpoints from our sample API.

   3. Fill the “Secret” field with any value. We can ignore this for now.

6. After clicking “Save”, the custom API should be automatically selected, and you can see the sample transcription result on the right.

7. That is all for creating and using Custom API! If you want to immediately use the API to create a project, you can do so by proceeding to the next steps to finish configuring and create the project.

Whisper ASR Limitation

Factor	Description
Accepted file formats	.wav, .mp3, .m4a
Maximum individual file size	25 MB
Request limit (Azure)	100 RPM
Request timeout	10 minutes
Maximum API call retry on request error (e.g. rate limit reached, request timeout)	2 retries