Audio restoration with very poor signal to noise ratio

Question

I am audio restoring and editing an mp3 file and converting to amr later on for distribution..see What are the best practices for sequencing the Isotope RX7 Standard toolset for dialogue audio restoration? for past history

Need help with following questions:

1) if signal/noise ratio is close to 1 (i.e. very high noise), is it advisable to amplify a lot prior to filtering or is there a better strategy?

2) The recording is essentially a monologue and sometimes the recording is so low that, I have to amplify a lot before being able to hear. However, even then, some of the peaks are at 15dB (which is very loud by any standard), and others are around 0dB. Yet, the signal is not heard strongly at max volume of speakers. Am I measuring wrong?

3) After final conversion from mp3 to amr file, the hissing noise always appears louder in amr file compared to the mp3 file. Is this an amr artifact?

4) What are the benefits of Izotope Dialogue Isolation offered in Izotope Advanced vs. the tools offered in RX7 Standard?

Answer 1

• Amplification of the digital signal won't make any difference to the de-noising process other than introduce the possibility of signal clipping, which will further degrade the signal. Recommend you don't do this. Just work with the original signal. Feel free to increase the monitoring level, but not the signal processing level.
• dB is not a absolute measurement unit - it only has relevance when referring to another dB level. When measuring signals you need to use it with a reference, such as dBFS
• Don't use AMR - it will degrade the signal too much.
• With Dialogue Isolate, you will just have to experiment with the trial version - every situation is different.