It was mentioned in the recent OggTraffic that Monty was working on Postfish but my impression was that it hasn't been released yet for public testing.

Then I googled a bit and found this page (in Japanese) with some screenshots of Postfish. So is it downloadable now?

From http://nyaochi.cocolog-nifty.com/audio/2003/12/postfish.html

It's in Xiph CVS, as the postfish module.

edit: You'll need GTK+-2.0 to build it.

It's my blog site. I built Postfish just for my curiosity although I didn't have any idea how to use it or what it is used for. I don't distribute Postfish binary because I think it's too experimental to use or evaluate it. I could build it on Fedora Core 1 without trouble.

Nice Time to get it from the CVS and have a little play

What is it?

Looks like it is some audio post-processing tool. I compiled it yesterday and had a play. It works on wave files and can output the processed wave. When you play the wave, there is a dynamic graph that shows the dB level of the left, right, etc. channels in different colours that may overlap. Very slick, in my opinion The only functional part, right now, is the declipping tool (shown in the second screenshot).

And, before anyone asks, it seems to be a Linux only version ATM.

Not necessarily. You do have gcc for windows plus I remember seeing a gtk for windows as well. Let me investigate

Oh, hi...

I've been working on Postfish very slowly behind other more important xiph.org work. It isn't released yet although I hope to make a nice prerelease for people to play with once I've coded in persistent state. Right now, it's rather frustrating to have n-thousand knobs revert to default state every time you start the app :-)

At this point, the version at svn.xiph.org has *almost* everything else implemented. All the effects work and most of them are intended to set a 'gold standard' for output quality. Aside from saving state, the missing pieces are the output config panel, the cuelist panel and the settings bank panel. I also plan a few changes to allow more flexible input options (mainly being able to combine multiple input files into channel groups rather than just chaining multiple inputs front-to-back).

The code should be portable, and compiling on windows (hopefully) is only a task of writing a Windows audio backend to go alongside the Linux OSS sound output. Same for MacOS X. The rest of the code is straight C99 relying on FFTW3 and Gtk2.

Monty

A newbie guide on what it is and how to use it would be great to have too in the future. I just downloaded Postfish from svn.xiph.org and managed to compile it. Other than playing audio, I've absolutely no clue how to do anything else. Pressing certain buttons does nothing.

Naturally, it will need a large user guide and tutorial, especially for beginners to audio processing. It *is* written to be as friendly to beginners as old salts. However, beginners will also need alot of tutorial in basic audio mastering and mixing techniques to make heads or tails of the functionality, whether they know what the buttons technically do or not.

Experienced audio engineers will be able to pick it up and use it but will naturally want all the painful details on what exactly it's doing at any stage.

For now I'm going to be tossing it out to the experienced folks to look for bugs; Xiph.org's dev schedule probably won't let me start a tutorial or a real release until end of June at the earliest.

Monty

The Postfish has been released by Monty!! Here is the announcement thread

I'd like a quick tutorial on how to de-clip a large WAV file.
I know it's gotta be a simple GUI thing, but i'm in the dark right now, and I have to deliver on a set recording that I screwed up =(

by the way, I came across this program just as the need arose... thank God for Free software developers!

I tried
$ postfish Track1.wav > Track1.new.wav

Track1.wav is 266 megabytes, Track1.new.wav is 680 megabytes
Then I made declip active, and pressed play. I waited until it reached the end of the file, and then tried to play Track1.wav in xmms - nothing... just skips to the next file in the playlist.
I will continue to blindly flip switches and push buttons until something works, or until someone posts the right answer =)

The default output of Postfish is 24 bit wav, which XMMS will not play. To get 16 bit WAV output, pop the 'ouput' menu and change the output bit depth from 'auto' to '16'.

It also looks like you fed it mono? Default output is 2 channel, so you probably want to set that in the output menu too.

Monty

Hello Monty;
(and all fellow HA members)

I'm not sure if this older thread is the best place to ask -- but I could really use some help... I've seen and read your "Postfish" announcement thread, but that one seems older than your last reply here - so here goes;

I've manged to d/l and compile postfish successfully, and it works!! (after much consternation getting all the various components of 'gtk+-2.x installed properly). My issue is related to trying to 'clean-up' (remove massive white noise, perhaps some pink too) from some (voice-only) mp3 recordings (all are 'mono') I have.

I have successfully converted mp3 to WAV, using Audacity, and fired up Postfish, and played (fiddled) around a bit with it's settings -- trying briefly 'declipper' and 'compander' (multi and single mode) which seems to change the sound a bit, and inherently remove 'some' noise, but not near as much as using 'Noise Analysis' and 'Noise Reduction' filters (Audacity w/LAME mp3 plugin on Debian Sid, or Nero's WavEditor in win98) seem to yield.

Can you please enlighten me on how to achieve better results ?
I'm obviously searching for a way to reduce the noise, even *better* than the atypical 'noise-reduction' filters can do...(they always sound 'flanged/phased' with the more noise reduction that is applied).

Perhpas "Postfish" isn;t the appropriate app to use for this -- but nonetheless, I've managed to compile form scratch and such (1st timer ;-) ), so it's been a very educational exp for me anyway -- and I'm sure to use postfish for *other* Audio projects in the future...

Thanks for any help or pointers yourself and any others can give me.

and a HUGE thanks for all your hard work, and the cdparanoia app, and ofcourse Postfish -- Excellent!! Bravo!!

Regards

Postfish is a nifty little tool. Having looked at it before I am pretty sure it was designed to act as multiband compressor. A lot of mastering engineers typically use those during the post production process. It's not like Audacity if my memory serves me correctly. Companding is basically compressing a waveform in order to send it over a channel, well in engineering at least and then reconstructing the signal in order to reduce crosstalk and noise. It's similiar to preserving the dynamic range.



The "phasing" tends to be an issue related to the FFT size. I understand where you are coming from with that as you can only remove so much noise.

Hi there and Thanks for replying;
something very odd about this forum -- I've been checking most every day to see if anyone replied to my query and TODAY July 27 is the 1st time Im noticing your reply/post. What gives ? The "quote" button doesn't seem to work either - so I had to hit "reply" -- hrmmm and even the [ quote ] Tags themseleves are not working(showing correctly) for me). I'm on Debian 3.1r2 Sid/Unstable and Konqueror as the browser (needed to move up this high to install Postfish (and it's necessary libs/tool set to build GTK+)). I should try again after this (viewing) using Mozilla and/or Epiphany and Dillo, etc



I hear ya -- and no it's not like audacity..


I was taught through some very old books that I barely studied but used as a reference years ago - namely "Recording Studio Handbook" (John M.Woram) that Companding is COMpressing and then ExPANDING -- I would guess similar to some of what Dolby does....but I've lost my terminology for these things. I had temporarily been interested in it when looking for a way to tweak/alter my sound while performing Live with a guitar... I happen to be more of a musician, rather than a Studio engineer, but I'm fairly technically inclined. On the other hand, I've slowly drifted(sadly) away from it all over the past decade or so (Family reasons). But my heart is still in it - and I'm slowly making a comeback (trying to anyways!).



I see (FFT) -- atleast I have some information about in which direction to follow up on concerning that flanging effect that occurs.

I immediately sensed "something" about Postfish, when reading here and other documentation, that the FFT had some connection to this issue, and thought great! I have an app (Postfish) that I can fiddle with (edit/alter) the FFT settings (perhaps able to tweak it) just ''so" -- So as to be able to kinda customize a multi-band pass-through tool just right, to enable removal of ONLY certain types of noise and frequencies (dynamically or not). Tweakable till I hear what I need to hear Rather than a cookie-cutter approach to noise-reduction. But that's only one aspect I'd like to get to figure out how to implement using this tool.
Much more playing with it is necessary (on my part).

Thanks again for replying

Edited late to fix broken "quote" tags ;-)

Make sure you have HTML activated on the board under your preferences. That's very strange.



Yeah it's fun stuff. Check out Ableton Live if you get the chance. If you have guitar and pre-amp you can connect them to the sound-card and record via line-in. Then with Ableton Or Audacity you can can apply some DSP effects, etc.



At my University you can earn a degree in Sound Recording Technology. The kicker though is that it's about 60% composition work on one instrument and the other 40% on the engineering. It's a Bachelor of Music degree though despite the fact. It's very difficult to find people who enjoy both. Normally most people enjoy composition side more than the actual technical stuff or vice versa.

Does someone know if there is a win build in the meantime? It looks interesting.

No it needs to be compiled from the source on Linux. . It's not well documented either.

Yes indeed --

[ removed all the borked web site stuff which is totally unrelated to this topic ]



I will do -- I've done so previously using a 12string acoustic, Live Mic (crappy), and Nero's WaveEditor -- just for kicks and giggles - not too shabby for what I had thought would sound just aweful, ...and useful for quick takes and just getting _something_ (an idea) down on tape(well digital), which can be easily toyed with (transposed, EQed, sampled, reverb, echo, flange, etc) ...just to toy with

[ removed more unnecessary borked website discussion ]



That actually sounds awesome to me -- wish that was available when I was younger ;-) though I'm not (wasn't) the college type then -- I was too busy being an "angry-young-man"...and doing things "My way"...

Regards -- thanks for the replies

[ Final bws removal ]