DASINU: a lot of information about recording music, part 2.

Recording part 2

In this part (part 2) the following subjects:
The soundcard (and the TC-electronic Konnekt8)
Make a loop
Mixing
Mastering

In part 1 you will find: introduction, development, Sequencers and recording software, from idea to song, Cubase SX3 / Cubase 5.

Mixing.

When finally all instruments are recorded a difficult part of the recording process arrives: mixing. Balancing the sound level of all the instruments and the sound color is a tricky business. For one reason because maybe you don't have professional studio monitors but also because mixing needs good ears, patience, experience and knowledge of the sound spectrum.
So, how can we achieve a reasonable result? My way of working is to first make a rough mix and work further from this. If you use (only) a headphone, use a good one. I use a Sennheiser HD265 that produces a "linear" sound spectrum. So, no emphasis on highs or lows. As for monitors, a lot of types are available on the market that are excellently suited for the homestudio (compact, active - so amplifier built in, nearfield and expensive...). I use the highly rated Genelec 8030A.

Genelec 8030A, excellent active nearfield monitors.

Lets start: the first thing I do is mute all parts/instruments except the bass and the drums. This is the basis of your song and it has to sound absolutely right. Except for the balance between the drums and the bass, you should take good care that the bass drum and the bass guitar do not peak in the same frequencies because this will generate a muddy sound. A parametric or graphic equalizer is an excellent tool to get a grip on this. Compress the bass and bassdrum (snare, toms etc)... it has to sound punchy (well, that's what I think).
Next step. To get a good overview of what I'm doing I make sub-mixes of e.g. the guitar parts. In Cubase you can route all channels of which you want to make a sub-mix to a 'group', E.g. all guitar channels to 'group guitar''. All drums to 'group drums', all synths to 'group synths' etc. These channels will all be routed as stereo channels to the main fader/strip. When you do this for clusters of instruments the mixing will be a lot simpler. Besides that you can now add effects (reverb, stereo enhancer etc) to a whole group, and that makes life easier.

When you are satisfied with the mix on your headphones and monitors you can start with the next process: mastering.

When you start the mixing process in programs like Cubase you will probably notice that equalizers, reverbs and compressors use a lot of cpu power. Together with the playing of the (say) 30 tracks you recorded this can result in a computer that's becoming overloaded, resulting in all kinds of unwanted side-effects (like short missing parts in your mixed wav file or stuttering). This can even happen when you have all your tracks in a 'frozen' condition, in which the plug-ins are not active and there output has been converted to a wav file. Experts tell me that even if your computer is managing your mixing process, a maximal use of your cpu has a detrimental effect on the sound quality. Equalizers are less accurate and the 'tails' of the reverbs don't sound so well. To solve this problem you do the following. You make sure that all instruments are balanced properly and you don't take care of the output volume and overall sound color. So you don't use any limiters or maximizers (at the general output)! Then you mix your whole song in 5 to 10 blocks, resulting in 5 to 10 wav files. By example in block 1 you mix all all drums and percussion to a file, in block 2 all the basses, in block 3 all the guitars and so on (use your common sense to structure the blocks, use the groups mentioned before). Take good care that none of the blocks are clipping at any moment of the track!.
After this you start a new project and import all the files in audio tracks. Mix these again, and at this point you use overall equalizing and the stereo field adjustment and maybe some compression (if needed).

hard rock mix
Hardrock mix.

Mastering

The purpose of Mastering is twofold. First of all this process is there to make sure that all the tracks you produced have the same overall volume and sound. Furthermore it's meant to polish the sound of your track. Just compare a commercial cd with the track you just mixed. In 99 of the 100 cases you will notice that your track sounds softer and more thin.
So how to master? First you export your mix to a wav file. I always do that in a 32bits format. Import this file in e.g. Cubase or a dedicated mastering program such as Steinberg's WaveLab. In the main channel you add a number of plug-ins. For most situations you need the following effect chain (and in this order):
- A bass roll-off filter (you can also do this with an equaliser) to remove all the low rumble and DC offset (let's say everything below 20 - 30 Herz). There are the frequencies you will not hear anymore but are using 'sound energy', meaning that the will trigger your amplifier and will try to move your speakers. Useless, remove those frequencies.
- A plug-in to enhance the lower tones. For example you can use Rbass and MaxBass that are in the Steinberg Waves bundle. They will give your low end more 'fat'.
- A plug-in to enhance the higher tones, to make them more crisp and clear. Plug-ins like that are know as exciters. The Spectralizer of Steinberg or the Sonic Maximizer of BBE are examples of that.
- A broadband equaliser to create the 'sound color' you want.
- A compressor, maximixer or limiter to increase the perceived volume of your track (loudness). Steinberg's 'waves' bundle contains some very good ones. A limiter will amplify the loudest part to just below distortion level and the lower parts adjustable more then proportional which can result in a very compact sound. With this method you have to take care that you don't remove the dynamics from your song, because if you do this in an extreme way everything might get the same volume level and you might notive pumping and sucking sounds caused by your compressor. How loud should a track be? Well, this is a matter of personal preference. But I would recommend you don't try to achieve the monster volumes of let's say Metallica. The last 10 years commercial cd's are getting increasingly louder. Look on the internet for 'loudness war', and you will see what I mean.
- To finish you need software to bring your 32 bit mix to a 16 bit format, the standard for commercial cd's and mp3's. For this you need a dithering plug-in. Often this functionality is already included in the limiter/maximizer.

Very handy is a tool like Ozone Izotope. It has all the functions above in one package and beisides that it is very easy to use,

When you have made your master play it on your home stereo. But before you do this, play a song of a professional produced cd in the same style as your new number one hit. After this, listen to your own song... yeap... sounds lousy huh? So, back to the drawing board. Most of the time the problem is not in the balance between the instruments, although this can be an issue also. But the amount of "lows" (too much or not enough - not enough punch, muddy) or - you guessed it - the amount of "highs" (too shrill or not transparent - not brilliant). Also the average volume level of the song can still be too low.... and a final possibility is that everything sounds like shit.....:-)
The next steps are obvious: the same procedure again until you have the right result. By the way, don't do this for hours in row, after some time you will suffer "ear fatigue" resulting in very strange mixes (when you listen to it later). As a final check (but I also do this in between masters) I put the song on my Ipod and listen to it in between other songs. I use this Ipod a lot so I know exactly what sounds good or lousy: it's my point of reference.

Mastering is a difficult job. Myself I'm never fully satisfied with the end-result.
But check for yourself, samples of my music are overhere. And I welcome tips, comments and feedback, for this you can mail me here.

Hans Soeteman.

Top of page
or
go back to part 1.

Site map:
Home (Dutch), Home (English), Over Dasinu, About Dasinu, Muziek, Music, Opnemen deel 1, Recording part 1, Opnemen deel 2, Recording part 2, Gitaar PodXt Patches, Guitar PodXt Patches (English), Apparatuur, Equipment, Biografie, Biography (English), Links, Links (English).

This page in Dutch

Links:

Line6. Modeling equipment
Cubase. Professional recording software
Genelec. Monitors
Reason. Website of Propellerhead, the creator of Reason
Reason manual. Useful site with a nice Reason tutorial (in dutch)
Motu: Soundcards
Konnekt8 My favourite soundcard.

Make a loop.

What you will do on a regular basis when you record music, is creating loops. For example a riff that should be recorded (and played) 8 times in a row can be recorded as 8 successive riffs in one wav file (say eight bars). But another way is to record one riff (or record it n times and pick the best one) and loop it eight times. You can now use it as many times you want. In this way you are not only more flexible, most of the time it also give a more "tight" end result. To cut a riff is simple, cut it at the beginning and the end of the riff, almost always on the beat or on the bar. A problem that now can arise is that when you play the "loop" you hear a very annoying click where the sample ends (and the new one starts). Something like...riff-click-riff-click-riff. This click is generated cause of a difference in the sound level at the end of the sample and the beginning of the next sample. Luckily there are methods to remove these clicks .A usefull and easy one is to let the software do it. Most recording software can find the so called zero-crossings in your sample and you can 'order' them (preferences) to cut only there. Zero-crossings are positions in the sample that are exactly 0db, so where they cross the red line in the middle in the picture below (which is a zoom-in to sample level of a wav file).

zero-crossing

When you let the software make the cut at a zero-crossing you can sometimes generate a new problem: although in most cases only a couple of milliseconds of your sample is removed it doesn't have that nice length of (say) exactly 1 bar anymore (but it will seldom be a problem). Another practical solution (it took me a lot of time to find it) is to make your own zero-crossings...how? Simple: at the beginning of the sample you make a fade-in of one or two milliseconds, and at the end a fade-out of the same length. It's so short that you cannot hear it, and your click is removed. Another way to fix clicks is to go to the edit mode of e.g. Cubase. Zoom into your wav to sample level and use the 'pen' to bring your sampl elevel to zero db just after and before the cut.
Top of page.

fade in
A fade-in of 2 millisecs

CoolEdit, modest but very handy.

The EMU 0404 USB. Excellent specs and not expensive.

The soundcard.

A brief section about the soundcard. To keep it simple: there are three important things you need to know about a soundcard.
1) the sample rate of the soundcard
A standard soundcard handles 44100 samples a second (so the sample rate is 44.1Khz). By example cds are based on this standard. A more 'professional' soundcard can also manage formats of 48 Khz (used a lot in the movie industry) and even 96 Khz or 192 khz. Theoretically you should be able to create a better sound quality with a higher (then e.g. 44.1Khz) sample-rate, especially in the extreme high frequency range. But (look it up yourself on the internet) the general opinion is that the audible difference is extremely small, if audible at all. A further disadvantage of a high sample-rate is that the resulting files will be much bigger and that you will need more cpu-power to handle these files.

motu828mk2
Motu 828 MKII soundcard

2) The number of bits with which a sample is stored. Simple soundcards can manage 16 bits music files. A cd is based on this 16 bits format. More 'professional' soundcards can manage samples also in the 20, 24 or 32 bits format. The advantage of a higher bit format is that a sample is recorded/played (extremely) more accurate (so more close to the original analogue sound). For example 24 bits gives 16.777.216 possible levels to store information against 65.536 for the 16 bits format. In practice the result is that your music will have less digital distortion (clipping), that effects can be applied more accurate with a better sounding result and that you have more 'space' (headroom) in your recordings (I'm quoting the experts here). At the end, after the mastering stadium, your recording has to be converted to the 16 bit format because this is the standard format for 'normal' musicplayer devices. To do this is a right way you need 'dithering' software, or a dithering functionality of your recording software.
It's recommended that you always use the 32 bits format when recording, mixing or mastering. With 32 bits you have a tremendous amount of 'Headroom', meaning that loud signals that are generating a digital diistortion won't do this in the 32 bit format. Only in the very last phase of your project, when you are creating your final wav/mp3 file you dither the signal to the 16 bits format.

3) The latency of the soundcard.
When you use a standard soundcard you will notice that when you play a tone on your guitar (via the pc), you will actually hear this tone a bit later through your speakers or headphones. This is caused because the pc, software and soundcard need time to processand transport the sound. Especially the transport is responsible for latency. How does this happen: the transport of the bits is organised in packages. With little packages there will be a more frequent exchange of information between soundscard and PC as there is with big packages. As a result the latency when using small packages will be smaller. It's also important to know that as a side effect cpu/pc has to work much harder when using small packages. The package size is deteminated by the size of the buffer of your soundcard (you can adjust this in the configuration screen). A standard buffer size is 1024 or 512, resulting in a latency of approx 40 msec.
When you are recording latency is very irritating, it's practically impossible to stay in sync with the music you want to record with. Every soundcard has a latency but in very good soundcards this delay can be very small and almost inaudible (5 to 20 milliseconds). Alas, in reality this low latency is not always (almost never) achievable because of the tremendous cpu pressure. In a big project (big number of tracks and midi instruments) this low buffer / low latency will then result in stuttering, distortion and drop out of sound (so if you have problems with stuttering sounds and so on, adjusting the buffer size will probably solve the problem).
Luckely a good soundcard has a 'direct monitoring' function. This means that the sound of your guitar (with hardware effects of your soundcard - if any, or outboard effects such as stompboxes) is put directly - without delay - on you r speakers, together of course with the music you are recording to. There still is a delay cause of the recording process, but you cannot hear it anymore, it's handled in the background. So, in my opinion a home studio cannot do without a soundcard with 'Direct monitoring'.

Konnekt8

My TC-Electronic Konnekt8, a fantastic, simple and trusty soundcard.

TC-Electronic Konnekt8.
When I bought my first "professional" soundcard I thought it was important to buy one with a lot of possibilities. So, a logical choise was the MOTU 828 MKII, lots of in- and outputs, loads of routing possibilities etc etc. But after some time I found out that all those possibilities were only bothering me (and I never used them). Besides that I was not impressed by the stability of the drivers and not at all by the service of customer-support
As a lonely guitar player in a home studio I only record one instrument at the time. I want to do that in a simple, but perfect way. That's why I bought the TC-Electronic Konnekt8 and I'm very happy I did. I now only use the two knobs on the front panel. With these I can adjust in a very simple way the in- and output level and the balance between what's coming out of Cubase and the sound of my guitar.Works like a charm. Besides that the Konnekt8 and the drivers are very stable and the sound quality is excellent.

Top of page