vocal isolator?

[Jp7010]

okay, i know there are a few vocal remover plug-ins, but, are there any vocal isolator plug-ins? plug-ins that'll do the opposite of a vocal remover, and instead of taking out the vocals, take out everything else?
"What's the use of such a plug-in?!", well, i'd like to know what the artist is singing.. as odd as that may sound, lol....

[Twilightseer]

I don't thing such thing exists.

[scottyc]

Originally posted by Twilightseer
I don't thing such thing exists.

Or, as it would sound with a vocal isolator:

Twilightseer's vocals isolated
I don't think such a thing exists.

[Twilightseer]

Someone must definitely invent that, then.

[supaman]

Home
17. Editing Sound Files
Some WAV files will benefit from a bit of clean-up before they are encoded. The most common forms of clean-up are trimming silence from the ends of songs, removing unwanted noise and normalizing the volume so all songs will play at similar levels. More sophisticated users may want to add fades, equalization or dynamic range compression. These effects change the nature of the music, so they should be used sparingly unless you are mixing a recording of your own music.

Files must be uncompressed (PCM WAV or AIFF format) before they can be edited. If the song is already in MP3 format, you can convert it to WAV format, edit it, and then convert it back to MP3. Each time you do this you will lose fidelity. Files encoded at higher (192 kbps+) bit-rates will lose less fidelity during the decoding/re-encoding cycle than those encoded at lower bit-rates.

Note: A few programs, such as MP3 Trim, allow you to edit MP3 files directly. These programs are typically limited to just trimming silence and normalizing the volume.

Sound Editing Software
Cool Edit and Sound Forge are popular programs that can record and edit sound files. Both programs are available in professional and “lite” versions. The lite versions include the basic features for trimming silence, adding fades, normalizing the volume and removing noise. The professional versions offer high-end capabilities for recording engineers and are overkill for most users.

AudioCatalyst and MusicMatch versions 4.0 and higher can normalize the volume automatically when the file is ripped. AudioCatalyst can also automatically trim silence from the ends of each track, and MusicMatch 4.0 can automatically apply fade-ins and fade-outs. For more extensive editing, you should use specialized sound editing software, such as CoolEdit or Sound Forge.

Normalization
Many CDs do not use the full 96 dB dynamic range that’s available. This can result in songs from some CDs playing much louder than others, even at the same volume setting. Normalization corrects this by scanning the uncompressed audio file to determine the peak or average level and proportionally increasing or reducing the levels throughout the file to obtain the desired volume level.

Prerecorded CDs that were digitally remastered from analog tapes are more likely to require normalization than CDs that were originally digitally mastered, but there is no hard and fast rule. Normalization is often needed for WAV files created from records and tapes.

Some rippers and most sound editing programs include a normalization feature. The Audiograbber ripper has the most flexible normalization feature of all the programs I’ve used, although some people might find it too complicated. The normalization features in Sound Forge XP and Cool Edit 96 are very basic and require working with each file individually. The professional versions of Sound Forge and Cool Edit are much better, but they are expensive and neither is as flexible as Audiograbber.

AudioCatalyst, which is based on Audiograbber, has a simple normalization feature that allows you to normalize all files to a set level, or to normalize only the files where the peak level is lower or higher than the thresholds you specify. MusicMatch Version 4 has a very limited normalization feature that adjusts all tracks to a single level.

Most normalizers allow you to specify a percentage of the maximum possible level for the highest peak. The maximum level may be referred to as 1, 100% or 0 dB, depending on the software. A setting of 50% (or .5) would be the same as –6 dB, because each doubling or halving of the signal level represents a change of 6 dB. A value of 100% (0 dB) will normalize the volume so it covers the full dynamic range, so the highest peak will be at the maximum level. Values above 100% should not be used because this will cause clipping wherever the level exceeds 0 dB.

Generally, all songs on a prerecorded CD will be recorded at about the same level, so you can assume that if the level needs to be adjusted for one song, the same adjustment will be needed for all the songs.

Normalization Settings in AudioCatalyst


If you have a CD collection that contains a variety of music, there’s a good chance that some songs will sound louder than others even when normalized to the same level. This occurs because the average volume may be different even though the peak levels are similar. Other factors such as differences in frequency content (especially with higher frequencies) and recorded distortion (electric guitar effects, synthesizers, etc.) will affect the apparent loudness of a song.

Listening is the best way to judge the appropriate level, but it takes time to listen to every song and normalize each one individually. An approach that works for most CDs is to normalize all songs lower than 91% or higher than 98% to a 97% level. Table 27 shows the results of normalizing four songs using these settings.

Table 1 - Results of Normalization


Old Peak Level
New Peak Level

Song 1
96
Unchanged

Song 2
85
97

Song 3
100
97

Song 4
93
Unchanged




If you want more control over normalization, Audiograbber provides advanced settings based on either average or peak levels, along with an option for dynamic range compression. Normalizing based on average levels will make the playback levels more consistent. However, raising the average level can easily cause clipping.

Audiograbber can be set to automatically apply dynamic range compression all of the time, or only when it’s needed to avoid clipping. Audiograbber can also be set to not compress songs that are already highly compressed. (See Chapter 11, A Digital Audio Primer, for an explanation of the difference between dynamic range compression and “compaction” type compression.)

If you have MP3 files that have not been normalized, you can use a player like Winamp to store preset equalization settings for each song. With the “auto” mode of its equalizer enabled, Winamp will read the setting for the song when it’s is played and adjust the level accordingly. It’s still better to normalize the WAV file before you encode it, because most portable and dual/mode CD players can’t compensate for individual songs recorded at different levels.

Transitions Between Songs
Professional DJs and anyone who records tapes for their own listening pleasure understand the importance of having a smooth flow of music. Playlists eliminate the need to swap records and CDs, but the problem of transitions between songs still exists.

Transitions between some types of songs need to be handled differently. For background music it’s OK to have a few seconds of silence between songs, but for dance music it’s usually better to crossfade between songs with no silence in between.

Trimming Silence
Many songs have a few seconds of silence at either end. Trimming off this silence will make the files smaller and allow for a more continuous flow of music. In most cases, you should leave anywhere from ¼ to ½ second of silence at the end of each song, unless you are making a continuous mix of dance music, in which case you’ll want no silence.

Many songs have excessively long intros or trailing sections of music or vocals that can be removed. If you remove one of these sections you should add a fade-in or fade-out so the song does not start or stop abruptly.

DC Offset

A condition known as DC offset can occur in sound files that were recorded with improperly grounded sound cards. This problem is more common with low-end sound cards. DC offset forces the baseline of the audio signal to be offset from the centerline. You can determine if this is a problem on your system by recording a few seconds of silence and zooming in on the signal and checking to see if it’s centered. Most sound editing programs have filters that can correct a DC offset.









Fades
It usually sounds bad to have one song end abruptly and the next one start immediately, except for dance mixes where the songs have the same number of beats per minute. Fade-ins and fade-outs can be applied to the ends of the songs to provide a smoother transition, similar to the way a DJ would use a crossfade.

Fades can be created with a sound editing programs, but it’s important to remember that these will be permanently stored in the file. Winamp has a crossfade plug-in that works without modifying the file. It works fairly well if you play most of your music on a computer, but if you use a portable or dual-mode CD player, you will need to create the fades by editing the WAV files. (See the tutorial for Cool Edit 96 in Chapter 20, for instructions on trimming silence and adding fades to a WAV file.)

Playing a song that slowly fades in immediately after a song that slowly fades out may result in too long of a lull in the music. One way to avoid the lull is to steepen the fade-in and fade-out slopes. Most editors give you several ways to control the “envelope” of the slope. Usually, you highlight the section of the file where the fade is needed, then you either graphically adjust the slope of the fade, or specify the starting and ending volume levels. Most sound editing programs include preset fade envelopes and the ability to define and save custom envelopes.

If you splice files together to make a continuous dance mix, you may have a hard time getting the beats to match exactly. One way around this is to use a sound editor to reduce or increase the tempo of one file to match the other and then splice them together so the beats match. The tempo adjustment feature stretches or compresses the length of a song, which effectively changes the tempo. DJ mixer programs like those by VisioSonic have built-in features for matching tempos of songs.

Optimizing Audio for the Web
Sound editing software can also be used to convert digital audio to different formats and to optimize audio files for use on the Web as downloadable music or streaming audio.

Internet access is slow for many people. Currently, most people connect to the Internet via a 33.6 kbps or slower analog modem. Even with a 56k modem, users are lucky to achieve connection speeds of more than 48 kbps. If you want the broadest possible audience for your Web site, it’s best to assume a “lowest common denominator” connection speed of 28.8 kbps.

Downloadable Formats
Compressed formats like MP3 are a good choice for just about any type of downloadable music. Uncompressed formats can be used for very short clips but should not be used for full-length songs. The advantage of using an uncompressed format like PCM Wave is that most Web browsers will be able to play it without special software.

You can choose from many different formats when adding downloadable music to your Web site, but you should stick with popular formats as much as possible. Otherwise, you risk losing users who may not want to install yet another player to support some proprietary format. Table 28 lists some of the more common formats for downloadable music.

Table 2 - Common Downloadable Music Formats

Type
Extensions
Format

MP3
.mp3
MPEG Layer-III

MS Audio
.asf, .wma
Proprietary (Microsoft)

QuickTime
.qt
Proprietary (Apple Computer)

RealAudio
.ra, .ram
Proprietary (Real Networks)

WAV
.wav
PCM*











*WAV files can use other formats besides PCM.

Small Is Beautiful
Your goal with Web audio is to create the highest quality sound file, at the smallest possible size, in the most commonly readable format. You can reduce file sizes (and bandwidth requirements) of both compressed and uncompressed audio. Even if you plan on using a compressed format like MP3, it still makes sense to tweak the uncompressed audio file to make it smaller before it is encoded. This will result in a smaller encoded file, as well.

The type of material, and desired sound quality are the two main factors to consider in optimizing an audio file. For example, for sound effects and voice, the sampling rate (which determines the frequency response) and resolution don’t need to be as high as required for music. Table 29 shows different combinations of sampling rates, resolution and channels that are appropriate for various types of uncompressed audio.

Stereo or Mono?
Is stereo necessary for the type of audio you are using? Certainly it is, if you are working with CD-quality music. For short clips and voice, using mono will cut the file size in half. Mono is also fine for many sound effects and background music.

16 Bits or 8 bits?
You can reduce the resolution from 16 to 8 bits and cut the file size in half again, but the signal will have more distortion from quantization errors (because it cannot be recorded as precisely with fewer bits). The difference between 8-bit and 16-bit resolution will be more noticeable in complex music with a wide dynamic range. For voice and sound effects, 8-bit resolution is usually adequate.

Table 3 - Web Audio Optimization

Type of Audio
Sampling Rate
Resolution
Channels
File Size of

1-Minute Clip

CD Quality
44.1 kHz
16 Bits
Stereo
10.3MB

Music Clips
22.5 kHz
16 Bits
Mono
2.5MB

Sound Effects
22.5 kHz
8 Bits
Mono
1.25MB

Voice
11.25 kHz
8 Bits
Mono
630KB




Sampling Rate
CD audio is sampled at 44.1 kHz and can reproduce frequencies up to 20 kHz. Most people can’t hear frequencies above 16 kHz. For music on the Web, you could use a sampling rate of 22.5 kHz, and many people will not notice any difference when using typical computer speakers. For higher quality music, a 32 kHz sampling rate can be used in place of 44.1 kHz and many people will not be able to tell the difference, even with a good speaker system. For voice, you can reduce the sampling rate to 11.25 kHz and it will usually sound fine.

Streaming Audio
Streaming audio is optimized by the streaming server and is usually compressed to deliver a higher bit-rate over slow Internet connections. Some streaming systems, such as RealNetwork’s SureStream technology, automatically optimize the bit-rate of each stream to the speed of the user’s connection. Other systems may need to use a different streaming server for each bit-rate.

For streaming audio to work well, the speed of your Internet connection must be greater than the bit-rate of the sound file. The Internet is designed to send data in scattered bursts. Good audio playback requires audio data to be delivered continuously, at a constant rate.

To allow for network congestion, the bit-rate should be no more than two thirds the available bandwidth. For instance, 128 kbps is considered the minimum bit-rate for good quality MP3 files, but this is much higher than the bandwidth that any analog modem can deliver. A bit-rate of 15 to 20 Kbps would be more appropriate for a 28.8 or 33.6 modem.

Table 4 - Streaming Media Systems

Type
Primary Format
Developer

Windows Media Technologies
Active Streaming Format (ASF)
Microsoft

Icecast (open source)
MP3
The Icecast Team

QuickTime
QuickTime
Apple Computer

RealSystem
RealAudio
RealNetworks

SHOUTcast
MP3
Nullsoft






Modem Speed

A modem’s speed does not equal how fast you can move data over the Internet. Some of the capacity is used by communications overhead and error correction. Variable telephone line quality also has a big impact on actual upload and download speeds. It’s not unusual to achieve speeds of less than 80 percent of an analog modem’s rated capacity. High-speed connection technologies like ISDN and ASDL can operate closer to their rated speeds because they have much less overhead than analog modems.

To listen to streaming audio at 128 Kbps, even a dual channel ISDN connection would be just barely enough. A higher speed connection like a cable modem or an ADSL (Asynchronous Digital Subscriber Line) connection would be required.

For short promotional clips of music, you should consider offering more than one format so your site will appeal to a wider group of users. It is becoming more common to find sites offering audio clips in multiple formats, such as MS Audio, RealAudio and streaming MP3.

The major streaming media systems support multiple formats, including MP3. But if you only offer streaming audio in a proprietary format like RealAudio or ASF, you risk losing users who may prefer to use a player like Winamp to listen to streaming audio. Table 30 lists some of the more common streaming media systems.

For more information on integrating audio into a Web site check out the book Audio on the Web by Jeff Patterson and Ryan Melcher, listed in Appendix C, Recommended Reading.

Sound Editing Utilities
Programs for directly editing MP3 files, processing WAV files and creating HTML interfaces for MP3 CDs are described below.

MP3 Trim and Wave Trim
MP3 Trim and Wave Trim (www.jps.net/kyunghi/mp3encod.htm) by Jean Nicolle are handy for editing large batches of MP3 and WAV files. Both programs are available in shareware and professional versions.

MP3 Trim allows you to edit MP3 files without decoding them to WAV format. It can detect and remove digital silence and truncated frames to recover wasted space. Without MP3 Trim, you would have to decode the MP3 file to WAV format, trim the silence with a program like CoolEdit, and then convert it back to MP3.

Wave Trim scans the first and last 10 seconds of a WAV file and removes the digital silence and any bits of audio left over from other tracks. Wave Trim also can normalize the volume, and accept command-line parameters for batch processing. Batch processing allows you to process a lot of songs without opening and editing individual WAV files.

MPEG DJ GoWave
MPEG DJ GoWave from Xaudio (www.xaudio.com) is a program for decoding MP3 files into WAV format. This is only necessary if you need to normalize the level or otherwise edit the sound, and you do not have the original CD or WAV file.

MP3 Prepare
MP3 Prepare (http://aryhma.pspt.fi/download/mp3prepare.html) is a program for making html-based user interfaces for MP3 CDs It can automatically create an html page that lists all albums on the CD and a playlist that includes all the songs. MP3 Prepare can also create separate HTML pages for each album, and include cover art, song titles and play times. MP3 Prepare features a range of fully customizable graphics and html templates, and includes MP3 Browser, which is an Explorer-like interface for browsing MP3 files on your hard disk. MP3 Browser can also display and edit ID3 information and generate playlists.

Ray Gun
Ray Gun from Arboretum Systems (www.arboretum.com) is a shareware program for removing noise from WAV files and adjusting signal levels. It can also be used to record audio. Ray Gun is available in both Windows and Macintosh versions and can be used as a plug-in for recording programs like CoolEdit and Sound Forge that support DirectX.

"Most of the information from human voice is in the 500 Hz to 3500 Hz range but you will not have much luck trying to filter this out of a song (for instance) since there are many other signals in this range and transients (such as drums) contain energy across a fairly wide spectrum. Your best bet is to try to filter the main vocals (I assume this is what you are after) by using the stereo positioning of the vocal track. Most vocals are mixed equally onto the left and right track of a song whereas instruments are often mixed more to one side or other. A filter that removes any part of the signal that is not in both tracks can help to isolate the main vocals. Many circuits for this type of thing have been published in hobby electronics magazines. Don't know of anything commercial off-hand .... this technique is very similar to what stereos with a "karaoke" setting do."
that was off a good sight and a forum topic. saved me from typing.
http://humanities.uchicago.edu/music/computer.shtml
learning
http://www.dps.com/
they have a solution if you talk to them
http://database.at.northwestern.edu/...m?LocationID=2
lot of things here that can be used together to do what you want.
http://www.jps.net/gworsham/studio.htm
some very very nonrelated basic things too know.


im to lazy to give much more, you can spend a few grand from some equip made by korg (its the cheap best way) to do what you want with amature (almost studio craft)work.
or you can use a combanation of software in the list and filter it ALOT of times. rember your voice k-band.
or if your savy you can build one your self (hardware) from some sight you should be able to find on the net.
hope it helps its all im willing to do now sorry.

[NeoRenegade]

Audiocatalyst and MusicMatch suck llama ass. Generally you should stay away from any program that has the Fraunhofer speed-encoder or the Xing encoder.

[supaman]

the programs in that were not the point, i thought that it was very helpful in understanding somethings.
however i do agree with you.

[Adasha]

Christ, that has got to be the longest reply ever!

Professional studio boxes exist to isolate and process vocal frequencies in audio tracks, but usually they cost hundreds or even thousands of pounds. One example of its use is with the Beatles' 'Free as a Bird' where they pulled out Lennon's vocal from an old recording.

Software wise I don't think anything exists. If it does exist it probably costs more than the hardware, but I'd love to be proved wrong.

You'll never get a brilliant result with a normal track anyway, because if other instruments have the same frequency or harmonics they can't be separated. Removing a vocal usually sounds better because there are other instruments to hide the gap.

It's not quite the same effect but try fiddling around with the equalizer or adjust the pitch with one of the plugins. If you only want to hear the words and not do something professional you should be able to pull out the vocal at least a bit.