How to increase recorded audio quality using audacity?

Question

My friend already recorded an hour speech a few weeks ago using his smartphone. But unfortunately, he put his smartphone was too far from the speaker, so the speaker voices in the recorded audio was too small if compared to noise. Now I am trying to increase the quality of recorded audio using audacity. I already tried to amplify, but it seems not giving expected quality. I found various features in audacity but don't know exactly how to use it in my case. Anyone can give me some idea how to fix the quality using audacity would be appreciated. Or maybe using aonther recommended tools.

Answer 1

Quality of recorded audio (in the digital realm) is measured by captured frequencies (sample rate) and dynamic accuracy (Bit Depth). Once recorded, changing those wont improve the signal, only increase the file size. If your noise floor is too loud when compared to your voice then you need to reduce the noise. There are expensive programs that do this quite well (Like like RX3 mentioned above). Otherwise, you can increase the gain of the signal and use selective EQ to notch out a thin band of frequencies where the noise floor is dominant. It wont sound as good as intelligent noise reduction or spectral repair, but it can make the voice more audible. You can also boost the dominant EQ area of the voice to help with focus. Remember to experiment and keep an unprocessed version of the original recording.

Eugene S, I hate to disagree with you here but Compression is definitely NOT the answer to his issue. In fact, compression will do the exact opposite. The Lack of Dynamic range between the voice and the noise floor is a big part of the problem. Compression will bring the noise level closer to the voice. You could use a multiband compressor to reduce a selectable area where the noise is dominant, but other than that I cant imagine it would be a good idea.

Answer 2

I reckon that the standard approach to such issue will generally have 2 steps:

1. Maximize the dynamic range and thus create a substantial difference between the speaker volume and the noise. I'm not an Audacity guru and I'm not aware what tool exactly can do such job. From a quick Google search I found this.
2. Apply a threshold under which the volume will be dramatically reduced (or even muted). This called gating. I found this link related to Audacity usage.

Assuming that done correctly, you will get a speaker voice without noise during the silent parts of his speech (between the words). Now you can try to play with things like EQ or multiband compressor (which is effectively the same thing) and try to reduce the background noise a bit more during his speech.