Re: [SoX-users] Mixing in a repeating track into a song

[Wojciech K. <web...@wo...>]

robs wrote:

> No hard and fast guarantees on this one, but what I would try is as follows:
> 1) Apply replay-gain tags to the music (e.g. with vorbisgain).
> 2) Record the lector as loudly as you can, or use vol or compand to get it
> loud; don't apply replay-gain tags to the lector.
> 3) Use sox -m --replay-gain=track with compand to do the mixing
> 4) Play around with the compander parameters if necessary. e.g. instead of
> -70,-60,-20, you could try -80,-70,-10.

  I played around with the compander, but I guess -70,-60,-20 is good. 
-90,-60,-20 seemed ok as far as the lector is concerned, but I don't 
want to do anything to extreme - the compander already has an audible 
impact on the song, so I'll probably stick to -70,-60,-20 and add -v 0.5 
for the lector and -v 0.4 for the song - does that sound reasonable ?

  I still don't have the lector track that will be used in production, 
so it's hard to say if I'll actually need the replay gain magic, but if 
I do, I'll use it on the waves, not oggs/mp3s - I see there are some 
tools out there for this. Anyway, I'll tell the people recording this, 
that I want it as loud as possible and that will probably suffice.

  One thing that's left is the clipping that occurs when using the 
compander. The man page says that to avoid this I should use the vol 
effect and set it to 70% of what `sox file.wav -n stat -v` returns - is 
this correct ? But what value should I use when mixing two files 
together ? Min/max/avg ? And what about the -v's I added ? Should I 
remove them when using vol ?


cheers,
Wojtek