On Thu, Feb 14, 2013 at 8:14 AM, Fons Adriaensen <fons@linuxaudio.org> wrote:

On Thu, Feb 14, 2013 at 01:40:23PM +0100, Brendan Jones wrote:

> Just personal preference, but I think both of these have an
> unacceptable amount of hissing that is not present in the original.

The original had most of the signal above 2kHz or so almost gone, so
this had to be boosted. Tape hiss gets amplified in the same way (this
was a _casette_ recording). It's probably possible to remove some of
the hiss using an FFT-based multiband expander.

Which is pretty much exactly what I did. I used WaveARTs MRNoise for this purpose, I have access to Izotope RX2 as well, but given that the signal there is so weak, I didn't use that in a destructive manner, and I am finding I prefer wavearts for primarily high frequency noise. However since RX2 is a new acquisition for me I am still learning their strengths and weaknesses.

> Can you explain what you did exactly

Second order shelf filter, +12 dB or so at 4 kHz and above. There were
some faint traces of signal around 8 kHz, so a very narrow band around
that frequency was boosted a few dB more.

Heh it is very interesting to hear you describe it, as I didn't look at the signal analysis at all when I did mine, but also put in a fairly sharp lowpass filter at 12k as I didn't hear anything usable above that when I boosted. Now I am curious to go back and try out what you describe as well as look at the statistical analysis and see if it can be improved a bit more or not, I just don't have time these days sadly. This was a quick 15 minute project or so to see what I could do out of my own curiosity while I had some spare time before a meeting.

Seablade