The Labs.Com Sound Lab Raw Sound Studio

The Labs - Design & Functionality For The Net

$MyVersion: 1.007 - Thu May 13 16:49:51 EDT 1999 - kiwi$

THE MOST RECENT VERSION OF THIS DOCUMENT YOU FIND HERE.

1. Introduction
2. File Format
3. Programs
4. Download
5. Applications
6. Sample Usage
7. Resources

RawSoundStudio a small Linux sound-package handling sound-play/recording and few simple mechanism to analyze the content. For the sake of simplicity a new (yes, sorry) format was created called .rss (raw-sound-studio), it's very straight forward and very simple to parse with C or Perl.

Raw Sound Studio 2. File Format

The format is following:

• Lines with '#' at the beginning are comments
• The first non-comment line describes the format, all argument on one line:
  1. 'signed' or 'unsigned'
  2. speed (int) [samples per second]
  3. bits (int) [8, 16]
  4. stereo (int) [0 or 1]

# A line starting with '#' is an comment and is ignored

signed 22025 16 0

<binary data follows ...>

Raw Sound Studio 3. Programs

Following programs are available:

• rsrecord - record sound-file in a certain quality
• rsplay - play sound-file

• rscut - cut pieces out of sound-file
• rscat - cat multiple sound-file(s) into one
• rsstrip - write raw audio data
• rsdump - dump audio in data
• rsecho - apply echo
• rspitch - apply pitch shift
• rstora - convert to real-audio
• rsfft - dump fft of sound-file

• rsview - display sound-file (average & fft)
  PerlTK script

Call each program with option -help, ie. rsrecord -help:

rsrecord: [-speed speed] [-stereo 0|1] [-bits bits] [-signed] [-unsigned] [-interactive] [-time mins]

Raw Sound Studio 4. Download

raw-sound-studio.tar.gz (read README)

Supported platforms: FreeBSD, Linux

To install run

make; make install

it compiles and copies all binaries into /usr/local/bin/

Raw Sound Studio 5. Applications

Unless otherwise noted all applications accept '-' as input-file which means stdin, and give the output over stdout. This allows you to cascade a stream of sound through several applications.

RsRecord

rsrecord: [-speed speed] [-stereo 0|1] -bits bits [-signed] [-unsigned] [-time mins] [filename] speed in Hz (ie. 22050), stereo (0 or 1), bits (8, 16), default is 'signed' if filename is obmitted then the recording goes to file dump.rss, default is 22050Hz, 16 bits, stereo on

RsPlay

rsplay: [-start start] [-end end] sound-file ... start and end are in milliseconds, so 10,000 are 10 secs

RsCut

rscut: [-start start] [-end end] sound-file ... start and end are in milliseconds, so 10,000 are 10 secs

RsCat

rscat: sound-file1 sound-file2 ...

RsDump

rsdump: [-start start] [-end end] [-avg avg] sound-file ... start and end are in milliseconds, so 10,000 are 10 secs, avg are in milliseconds too, it builds the average (magnitude) over that period of time, normalized to 1.

RsEcho

rsecho: sound-file delay amplification delay in ms, amplification 0-1. (usuable value is .6)

RsPitch

rspitch: sound-file pitchshift pitch-shift 0.1-2. (usuable values are .8 ... 1.25)

RsView

rsview: [-start start] [-end end] [-avg avg] [-fft] sound-file ... start and end are in milliseconds, so 10,000 are 10 secs avg are in milliseconds too, it builds the average (magnitude) over that period of time, normalized to 1. rsview is a PerlTK-script, it uses rsdump and rsfft internally.  % rsview -start 25000 -end 40000 english-0.rss   % rsview -fft -start 25000 -end 40000 english-0.rss   % rsview -fft -start 29100 -end 29100 english-0.rss

RsToRA

rstora: [-quality 0-7] [-title title] [-author author] [-copy copyright] file.rss Converting rss-files to name.rm (not via stdout) realaudio files (highly compressed). Requires raencoder which you can download from Progressive Networks: Real.Com, look in area of content-creation tools. Note: rstora requires the same disk-space on the /tmp/ as the file you like to convert, for long audio this can be hundreds' of mega-bytes. We tried to pipe raw data to raencoder, but the resulting realaudio-file was broken all the times, writing the data into a file first and calling then raencoder works fine. At a later time when raencoder is fixed rstora won't use any additional disk-space just for converting but only as much as the resulting realaudio-file itself.

RsFFT

rsfft: [-start start] [-end end] sound-file ... > fft.list start and end are in milliseconds, so 10,000 are 10 secs

RstoMP3

rstomp3: file.rss converts file.rss to file.mp3 Download MP3 Encoder and compile it, install it and don't forget to setenv MPEGTABLES to the right directory. Then rename encoder to mp3encoder and copy it into /usr/local/bin (or alike dir), then rstomp3 is usable (it's a simple csh-script), it uses /tmp/ to create raw-file for mp3encoder.

MP3toRs

mp3tors: sound-file Converts .mp3 to .rss (stdout) Download MPG123 (MPEG Layer 1,2,3 decoder) and install it, then run mp3tors

Raw Sound Studio 6. Sample Usage

Record audio from live-input for an hour in CD-quality:

rsrecord -time 60 -stereo 1 -speed 44100 live.rss

Cut out a sample of 10 secs at position 0:45:

rscut -start 45000 -end 55000 live.rss > cut.rss

Apply double echo:

rsecho sample.rss 10 .5 | rseffect - 200 .6 | rsplay -

Convert raw sound-file to .rss:

echo "signed 16 44100 1" | cat - sound.raw > sound.rss

Convert MP3:

mp3tors art-of-noise.mp3 > art-of-noise.rss

Raw Sound Studio 7. Resources

Sound Lab Resource listing	Audio Programing Collection of links	Audio Effects Explained Description of common audio effects
Audio Format Overview Usefule file-format FAQ	4Front: Audio API Audio-lib for LINUX

                                                                                                                                   

MP3 Lab

Sound Lab

SoundArt