improving acoustic ...

improving acoustic quality  


New Member
Joined: 4 months ago
Posts: 1
01/07/2019 10:05 am  

Hello everybody.

I have a set of audio files (~1000 hours), all speech of a single speaker. A good portion is in reasonable quality but many have various acoustic defects. I thought I could train a neural network to learn what a good recording looks like and then translate the defect ones to good ones, much like we can use a CycleGAN to translate a sketch to a photograph. But the problem is that the defect recordings are not all defect in the same way. Some have a lot of noise, some have echo, some have high frequencies cut off etc. So I cannot really train a CycleGAN in the traditional manner because I don't have two domains between which to translate but a variety of domains and combinations among them.

I wonder how to tackle this problem. Any ideas are welcome.

Topic Tags

We use cookies to collect information about our website and how users interact with it. We’ll use this information solely to improve the site. You are agreeing to consent to our use of cookies if you click ‘OK’. All information we collect using cookies will be subject to and protected by our Privacy Policy, which you can view here.


Please Login or Register