Can we use data from other track?

Children’s Speech Recognition Challenge
1

I have two questions:

  1. Can we use audio from the Word track for the Phonetic track?

  2. Are we allowed to use the Real Class dataset for the Phonetic track, or is it only permitted for the Word track?

2

@dzunglt24 Yes, per the Problem Description, “Training data, including audio and transcripts, can be used across tracks.” This includes the Real Class dataset. Thanks for the question!

1 Like
3

Only the subset that you provided, right? Also, can we use external noise datasets (not associated with the ones in the prohibited list) for broader coverage?

4

I’m not sure what you mean by subset. I’m referring to the provided training data in both tracks. You are allowed to use external data as long as it meets external data requirements. Please refer to the home page and rules for more details.

6

@cszc Kindly confirm if CC BY 4.0 is allowed under your license rules.

7

Yes, CC BY 4.0 is allowed. Please be prepared to provide proper attribution if asked.

8

@cszc Kindly also confirm if this license qualifies your license rule CC BY-SA 3.0.

9

Hello! To be prize eligible, you must be able to share your model with an MIT license, and doing so must not violate the terms of any tools or data used.

Whether you could use share-alike licensed materials in developing your solution and still be prize eligible depends on what the materials are and how you are using them. We encourage you to review the relevant license terms carefully to determine whether your intended use would be compatible with challenge rules.