Is there a tool to identify the likely sections having an echo/reverb within a conversation audio?

I was suggested a similar topic, and I found a Nyquist tool at post no.12, “How to remove echo from recording?”.

From there I was led to Dereverb and the tool, Dereverb.ny.

The problem is that of the two speakers only one has a reverb/echo because probably of a tiny room.

Is there a way to automate the dereverb process that could identify only the sections that have reverb/echo?

It seems I have to manually select the sections that has the reverb, presently.

Is there a tool that can identify which audio sections most likely have a reverb from a full audio file? I used a particular conversation from YouTube, which might help put the matter in the right perspective.

Sounds like a job for AI, e.g. … Free online reverb remover tool powered by AI - Voice.ai
[ it only applies de-reverb where necessary ].

Wow! I thought so as well.
However, couldn’t this de-reverb issue be programmatically approached? Like comparing a small enough signal bit with its successor, multiply the earlier bit with a fraction dependent on the time it takes for the parent sound bit to bounce back from the surrounding obstacles, then subtract it from the successor, then observe how sharp the signal becomes? This step has to be repeated a few times internally until the fraction is determined for a particular enclosure.
Yes, there are added complications involved in this programmatic approach: Different frequency bands of the constituent components of a sound wave are generally absorbed and reflected differently. Also, the velocity of the constituent frequency bands of a given sound wave in air isn’t strictly independent of these bands. Very complicated scenario indeed to be solved by one individual in one sitting.
But I thought that could surely be a group of outstanding acoustical physicists and engineers who would definitely have had a solution by now to almost completely have included all the variables and have solved the problem.

In the end, a big thank you for the link.

This topic was automatically closed after 30 days. New replies are no longer allowed.