I’m my sister’s audio editor for a startup audiobook narration business she and I are just beginning and I am looking for a feature that would let you select a portion of the waveform, have Audacity recognize it and find all similar waveforms in length and amplitude so they can be either replaced by room noise floor sound or deleted altogether as you can do in Word or other similar programs. For example, it would be great for removing all mouth sounds (or even if it worked on half of the instances, it would be a great time saver for the audio editor/engineer) from a recording or singing composition instead of the current way of finding each and manually removing each one. If there is something already in Audacity that is like this and I’ve missed it, please let me know. I’ve been looking since our first recording and haven’t found this feature.
have Audacity recognize it and find all similar waveforms in length and amplitude so they can be either replaced by room noise floor sound or deleted altogether
You’re describing Noise Reduction. The Profile step lets you point to a sound segment that you don’t want (let it sniff the sound), then you actually run Noise Reduction on the whole show and it takes out or suppresses similar sounds.
Note sound is nothing like a spreadsheet. There is no global delete the letter “A”. It’s more like look at this small patch of grass in your lawn and take out all similar patches of grass.
it would be great for removing all mouth sounds
Did ACX actually complain about mouth sounds? I don’t know that they ever complained about natural breathing and human noises. If you have a microphone that produces “Essing,” (gritty, strident SS sounds) you can run one of the DeEsser programs and suppress that effect through the whole show while you go make coffee.
There is also a DeClicker in the same group.
We also have an Audiobook Mastering suite that’s been pretty effective.
Let us know.
I don’t normally ask for this in a programming question, but it would be interesting to see what happens.
Record a 20 voice second test and post it on the forum. Don’t Process Anything. Raw sound.
Instructions in the panel links.
While searching based on length and peak amplitude would not be very difficult to implement, it would be hopelessly inaccurate.
For accurate matching, you also need to look at the frequency content, and how the sound changes over time, and then look for close matches (exact matches will virtually never occur except in the case of synthesized sounds). This is extremely difficult to do in software, and Audacity does not have this ability. Developing this feature would be a major software development effort, and is unlikely to happen in Audacity in the foreseeable future.