rosea grammostola wrote: yes, in the /convolver/new command. in the /impulse/read command, it's used to specify *which* in and out. (e.g. there can only be one of each, per line.) yes, and no. the minimal size is jackd's period size (if you specify a smaller value, jconv will tell you it's using a value equal to jackd's period). you can use larger values for more cpu efficiency, at the cost of higher latency. not quite. it's the amount of space that jconv allocates, i.e. don't make it much larger than the largest file you intend to use in that config, it's wasted. you can use the offset to cut away the impulse of the direct sound, for aux-send style reverb. cutting the length could make sense if you want to isolate just the early reflections and omit the reverb tail. delay is interesting if you want the reverb tail treated separately, because you must delay it to set in after the early reflections. check out the example configurations for some rather involved setups that demonstrate what can be done. nothing prevents you from using aliki to capture some old fender amps. but you should take into account that it only makes sense for the linear operation, i.e. the tone color of the clean amp and speakers. as soon as distortion sets in, simple IR will no longer cut it. my choice of reverb does not depend on the instrument, but on the setting. personally, i hate mixing different rooms in a single mix (the old rocknroll disease of using a special snare reverb, another one for the rest of the kit and a plate for the vocals). i choose IRs mainly for decay (do i want to recreate an intimate, small venue or a concert hall/church) and color (hard, glass walls, or wooden panelling etc.), and then i stick to one IR for all instruments, varying the dry-to-reverb ratio to simulate depth. only a few days ago have i started to experiment with multiple ambisonic IRs (one per instrument position), but so far i've screwed up. takes a lot of practice and patience, and it's very easy to mess things up.

On Sun, Sep 27, 2009 at 10:59:06AM +0200, rosea grammostola wrote: It's not a reverb, but a general-purpose convolution engine - which of course can be used for reverb. Seems you didn't read README.CONFIG... Please do, let me know if anything isn't clear. The parameters in jconv's config are there mainly to allow you to use many IR files without first having to modify them. That in turn is because I can't provide 'prepared' IRs in many cases, either the data files are very big, or their license does not allow me to distribute them. But the effect is mostly determined by the impulse response itself. So there really are no 'typical jazz' parameters... Depends on a lot of things, including the type of music and the type of sound you want to create in a mix. There are basically two reasons to use reverb: 1. To create a 'natural' sound, i.e. one that includes the acoustics of a real space, or something that could be a real space. In most cases, if the 'real space' is not something special such as a church, the listener would not really be aware of the reverb and certainly not hear it as an effect. It would just add realism, provide a idea of the dimensions of the space, and create depth - some instruments being closer than others. This is what you would do for classical music and in general for anything called 'acoustic'. In that case, if you start with dry recordings, you would add reverb on *all* instruments and voices, but not the same amount on all. 2. As an effect, mostly applied to a single voice or instrument. Here anything goes, and you can use types of reverb that don't correspond to any real space or that even defeat the laws of physics, e.g. a reverb that cuts off before it has fully decayed, delayed reverbs, very dense or heavily filtered ones, etc. There are also two ways to 'wire' a reverb unit into a mixer. A. The 'traditional' way (from the analog multitrack days) is to use a post-fader auxiliary send on each channel to send a controlled amount of signal to the reverb. The signal from the reverb is then mixed in just as any stereo track would. In that case the reverb should be 100% 'wet' as the dry part follows its normal way through the mixer. For an IR reverb you need to cut of the direct sound, as is done in most examples that come with jconv (that's mainly why the offset parameter is used for). This method allows to share a reverb among many channels, while still allowing to control each of them individually. It is what you would do for use case 1. above. When using a digital mixer you could also use small delays on either the dry sound or the reverb send to enhance the sense of depth (that's less traditional, as analog mixers couldn't do this). B. The second way more corresponds to use case 2 above, just use a reverb as an insert on a single channel. In that case you need a different one for each channel. And of course both methods can be and often are combined. Ciao, -- FA Io lo dico sempre: l'Italia è troppo stretta e lunga.