Will your card be used with a PC, a laptop or both?
Soundcards
can interface with your computer internally on a PCI slot, on laptop
via PCMCIA slot, using Firewire 1394 or USB. If you want to use you
card on both your Laptop and PC, Firewire or USB might be the better
choice for reasons of compatibility and portablility. The installation
is also usually
very straightforward.
The advantage of a PCI card is that it connects
directly to the computer using the PCI bus. This bus is faster
than Firewire and USB, where an additional component is present between
the
soundcard and the processor. Even if the USB or Firewire
adapter is phsically on the motherboard, they are connected via the PCI
bus. This will add a small latency (delay) to the incoming and outgoing
audio. The CPU
usage of devices on such a buss is also slightly higher.The additional
delay is only in th region of a few milliseconds, but after aother
delays are considered, this additional delay may make the difference
between "noticable" and "annoying".
Are you likely to use a mixing desk in your recording setup,
or would you prefer the setup to be as small and portable as possible?
Soundcards come in many shapes and sizes, some with the basic ability
to receive 2 line inputs and 2 line outputs, some with microphone
inputs with phantom power, instrument inputs, MIDI connections,
Headphone outputs, ADAT, Optical, SPDIF, and clock. It is wise to get a
card which will accomodate your present and future needs. It may also
be woth considering getting a device with everything you will
need i one box. There are now an increasing number of products
available which include a control surface with flying faders and a
mixing desk.
Some people think that a very simple card will satisfy their needs and indeed this might be the case, but many people overlook the need for good monitoring and the ability to record more than one stereo track at a time when they select the model. Another common mistake is the belief that more than one sound card can be accessed by one music production program. This usually not the case. In professional audio soundcards use ASIO drivers (Audio Stream Input/Output). This type of driver allows the card to input and output audio with very little delay (latency) This is very important for recording music because when a musician records lots of subsequent tracks they should be synchronized, something which we took for granted in the days of analog recording because the tracks were all recorded in parallel stripes along a piece of tape. We also need to monitor the sounds we have already recorded with no apparent delay. Otherwise subsequent tracks recorded will appear to have been recorded late. This can often be achieved by fine tuning the way that the soundcard communicates with the computer, but ASIO soundcards offer another solution. The better cards contain a mixer which will allow the user to mix signals between the inputs and outputs without the delay associated with latency. This way it is possible to create "no latency" mixes for monitoring whilst we record. Some of the less expensive cards do not allow accommodate this option, whilst others do and the capabilities of the internal mixers vary. RME Totalmix and MOTU Cuemix are very good examples of mixers which are controlled from the computer, but actually take place in the card itself. Beware, some of the ASIO drivers provided with cheaper cards are inadequate to say the least. Spend some time checking the reliability of the drivers and the capabilities of the internal mixer before you commit to spending your money.
Recording on computer configured as a DAW (digital audio workstation) has become more accepted in recent years, indeed many of the people I encounter now have never seen an analog recording setup. With the popularity of this shift in the recording industry has come an influx of far cheaper products from unscrupulous manufacturers. BEWARE. Often a manufacturer will market a consumer product with a professional badge on it rather than producing a good product. Marketing on the basis of numbers which people do not understand is an established method of extracting money from the un-suspecting punter. Car "Hifi" definitions of power output demonstrate the sort of marketing to which I am referring:
A power amplifier for a car is advertised as having an output
of 600W PMPO. Sounds impressive doesn't it. Now let's look at the
detail. Hifi, Studio and P.A. amplifiers in the UK are rated using
Watts RMS (Root Mean Square) for a speaker impedance of 8 ohms, per
channel. PMPO is something more recently invented by the domestic audio
industry to impress the un-knowledgeable. It refers to Peak Music Power
Output for the whole system. So divide by 2 to get to British Peak
Power, Divide by 2 to get to unclipped RMS, Divide by 2 to get to the
output per speaker in a stereo system and divide by 2 to obtain the
output for 8 ohms per speaker instead of the 4 Ohms a car amplifier or
low quality domestic "Hifi" would expect to address: 600/2/2/2/2 =37.5W
RMS per channel approximately. Not quite so impressive, but far more
than you need for a small space like a car!
The truth is that the loudness of sound is not measured in Watts. The
unit of sound pressure level (SPL) is expressed in Decibels, The amount
of change in air pressure compared to a certain air pressure with no
sound present in it.
The other reality is that a 600W sound system would drive a car battery
flat in very little time!
We have to contend with slightly different marketing criteria.
Common examples are sound cards which are advertised as:
Windows' built in sound system "Windows Direct Media" or WDM was never designed to produce cutting edge, low latency, high quality results. Most built in soundcards on a motherboard are simply designed to play with the relative high fidelity of CDs and DVDs. They can do this well enough for domestic users, but if you are serious about making or recording music you will need a sound card supplied with an ASIO driver (Audio Stream Input/ Output). This is a technology developed by Steinberg Media Technology. The driver allows a soundcard to perform certain functions that are not possible with a WDM driver. Part of the ASIO specification requires that the card reports its true latency back to the audio application. The recoding app. can then compensate fro this latency to ensure a synchronous output from all its channels and tracks. This might seem somewhat complicated, but when you consider that it is quite normal to have a huge number of fragments of audio file scattered all over you hard drive. These must be re-compiled, buffered in memory and then output to your soundcard with sample accuracy. In itself a demanding task. Another function of the ASIO standard is to provide Direct Monitoring. This usually consists of a mixer associated with your sound card, but not the computer. This mixer allows you to create a dry mix to provide monitoring with no latency at all whilst you are recording. It is perhaps easier to understand as a separate monitor-mixer which completely avoids the latency associated with computer recording. Some applications for recording music allow you to control this mixer from inside the program. Without ASIO, direct monitoring is not possible unless you use a separate mixer to create a mix to monitor whilst you record.
All but the most basic music programs allow the user to choose the sample-rate and bit-depth settings. The latency/buffer settings are usually controlled from within a window peculiar to the soundcard, but accesible from your recording software. It is usually called "ASIO Control Panel" or something similar. Beware. Some domestic soundcards are advertised as having an ASIO driver, but the control panel does not allow the user to change these settings. You will often need to change them depending on the stage you are at within a recording project, to get the best functionality from both your computer and the soundcard. Often the Direct Monitoring facilities of the card are also controlled from this window.
In short, the minimum requirements of a soundcard for recording should be:
I cannot stress this too much. You should become conversant with the ASIO control panel and the Direct Monitoring facilities as soon as you can. Don't worry, the settings can be changed at will so long as they aren't changed whilst something is playing or recording. In most cases, because these settings are peculiar to the soundcard and not the recording software, your music program cannot save the soundcard settings. They will stay the same until you change them, so for example, switching to another program, the soundcard buffer/ latency settings will stay the same but the bit-depth and sample rate will change according to the program in use and the settings you have chosen for your project.
A brief note about ASIO:
DAW software can only output sound to one ASIO driver. This is a
limitation of ASIO, so much as you might have several soundcards
installed on a computer, you will only be able to use the sound devices associated with one ASIO driver at a time.
Some ASIO drivers are written by the manufacturer to synchronise with more than one soundcard. That is to say, you may be able to use multiple soundcards from the same manufacturer if their ASIO driver supports this function.
Even though the MIDI input and output may well reside in the same box, the MIDI driver is entirely separate and should be viewed as a separate device. It is quite normal to use multiple MIDI devices from within one DAW application, so you may well be able to treat a lesser soundcard as another MIDI input/output.
In the old days of recording, many tracks were recorded along side each other on a wide piece of tape. The tracks were locked together and precisely synchronized. To hear a track which had been previously recorded whist recording another along side it was a simple matter. There was no time delay between hearing one track and recording another.
In the computer domain this is one of the most important things to consider because it has great implications, not only in terms of if the computer will work, but also in terms of how the process has to be handled.
Computers can only perform a limited number of tasks at one time, so to perform multiple tasks that appear to happen together a computer stores completed tasks in memory and releases the tasks later, but synchronously. This delay is known as Latency. It occurs to a different degree in different devices within the computer and for different reasons. This is one very important reason to have a reasonable understanding of the devices in your computer (not just the sound devices) and which resources they need to make them work efficiently.
There is no real need to go into great depth about error correction at this stage, but it is very important that it gets a mention. You will seldom see error correction featured as a reason for a computer failing to do its job, but a lack of it is very often responsible for our audio sounding bad. you might think that a if computer says yes, it says yes. If it says no, it says no. Well hopefully that will be the case nearly all the time!
Each process in computing is handled by devices with many millions of switches changing their state at very high speeds. This includes the CPU, USB devices, network devices, sound and graphics cards, memory and all the components which allow them to communicate with the memory and processor. Often a device will fail to get data from one part of the machine to another, or to process it exactly as expected. This happens constantly in all computers, all the time and between all the devices connected to it.
This is where error correction comes in. Complicated computations are used to detect exactly what the correct data is supposed to be, then the data is passed from one device to another. The data is then held in a small area of system memory called a buffer and released as packets of information of a predictable duration in sequence. These packets of information are known as samples. Each time this happens a small delay occurs. If too little time is allowed for this process to take place, information may cease to be passed along the chain of components in a predictable manner. The samples are released irregularly and the sound becomes distorted, or if the buffer is too small to complete the tasks with some memory free, the output can cease completely, so:
Lower latency (smaller buffer size) = shorter delay but more likelihood of bad data
Higher latency (larger buffer size) = longer delay but better sound
We can instruct the program to use a larger or smaller amount of memory to process audio information going to and coming from the sound device. Here are a few examples:
High quality recording: 96Khz, 24bit, Latency (buffer size) set to 128 samples = delay of 2-3 milliseconds, very low delay, high probability of bad data, but only 4 devices in use, 2 to record and 2 to play back. I'm monitoring through the computer with very little delay, but the computer can kkep up comfortably because it is not doing much effect processing.
Live recording of 16 channels at the same time, 48khz, 24bit, Latency set to 1024 samples = delay of 22-23 milliseconds. Long and clearly audible delay, but the recording is live and doesn't need to be monitored instantly, so Latency can be traded off against accuracy. A large buffer size will ensure that the computer can keep up with the recording. In this case I'm also considering that the buffers need to be large enough to allow time for the hard drive to skip around the disk, recording little peices of 16 different traks in sequence.
Multi track Editing: 48khz, 24bit, Latency set to 256 samples. The delay is 6ms. I can edit all my audio files and hear it with only a slight delay, but I haven't added lots of effects yet, so the audio engine isn't straining the CPU.
Mastering: 96Khz, 24bit, Latency set to 1024ms. Monitoring doesn't need to be instant, but I want to add some very high quality processing to stereo tracks, so the high latency allows the audio engine to do its job with better data integrity and less chance of drop-outs. At this point I don't need to hear anything from the inputs, so the delay is not going to worry me. I'm only listening to the output from the computer.
Most sound cards also have a number of MIDI inputs and outputs, at least one. There were a number of MIDI adapters which didn't function very well years ago, but this problem is broadly speaking not a very important issue now unless you are considering a large MIDI setup, or are considering using MTC (Midi Time code) to synchronise. external equipment to your setup. Most MIDI interfaces work well. It is a good idea to choose an interface with at least one input so that you can control your music software with a keyboard. One of the great joys of a modern DAW is it's ability to host virtual instruments.
Do not be fooled into thinking that an average soundcard will do justice to your recordings. Inside the DAW you will perform many operations which can potentially damage the quality of each sound you record by the time your recording is finished. Recording with superior fidelity at the start of the process will impact your final output by the time you have mastered your recording. I cannot overstate this. A soundcard should represent the sound accurately. It should not flatter the sound in any way. Otherwise it will be very difficult to create a mix that will sound good on other systems than your own. Yor soundcard should be of "reference" quality, not "Hifi" quality.
This site is under on-going development. More pages will be added soon. Thanks for visiting inthepipeline.net