Independent ALSA and linux audio support site

Sound Cards: Introduction

From the ALSA wiki

Jump to: navigation, search

In the early days of personal computers the only sounds that the computers could produce were created by a small permanent-magnet loudspeaker connected to a simple tone generator. This facility was very limited, and afforded the programmer little more than the ability to produce a few 'beeps'.

To address the growing demand for more capable sound hardware, several manufacturers began to produce plug-in cards ('daughter boards') which contained more complex sound devices. The famous 'sound blaster' is one example. These plug-in cards were of course generally called 'sound cards' and, being 'daughter boards' they plugged into the computer's 'mother board'. Sound cards generally have no transducer (loudspeaker, microphone) capable of converting between electrical energy and sound energy. They rely on external devices, often connected through external amplifiers, to do that.

More recently, microelectronics has enabled greater integration, and it is now common to find fairly complex sound devices on the motherboard itself. Although these electronic devices are quite often not separate cards, they are still referred to as 'sound cards'.

In ALSA a soundcard can have several capabilities:

Different soundcards will have different hardware features. A basic soundcard might have the I/O ports shown in the figure. More sophisticated cards will support more channels, S/PDIF I/O, MIDI ports, or more analog speaker outputs for surround sound or woofers.


In the figure:

\ The 'Sound Card' unit usually consists of three sets of components:

The interface between the sound devices and the processor is usually through the computer's PCI (or PCIe) bus. The hardware which implements the interface is usually contained within the 'Southbridge', part of the 'chipset' produced by a manufacturer for motherboard designers. The name given to the 'Sound Card' is derived from any of the following:

The program alsamixer reports the names of both the 'sound card' and the 'audio chip'. As a result with a motherboard based on a NVidia chipset the output may be:

Card: NVidia CK804
Chip: Realtek ALC850 Rev 0

On another motherboard using a Via chipset, one might find:

Card: Via Via8235
Chip: Realtek ALC850 Rev 0

The audio CODEC chip, a Realtek ACL850, being the same in both cases. The low level command

cat /proc/asound/cards

provides the card name and chip number but leaves out the vendor name 'Realtek'. Unfortunately most high level programs only provide the card name.



The audio cards used in most home computers are based on the Intel AC'97 standard (see below). This defines a set of standard controls on the audio chip and is sufficient for high level programs to find out what controls are available on each card and to then issue the correct commands.

Unfortunately, the PCI interface is not fully specified by AC'97 and so different ALSA drivers are needed for each type of PCI interface. As a result, in a running system, the command

lsmod | grep snd

will show the module snd_ac97_codec which handles the AC'97 standard. It will show that this is calling a second module to handle the PCI interface. For the Via8235 above this would be the module snd_via82xx. For the Nvidia CK804 it will be snd_intel8x0, as Nvidia and Intel use the same interface standard.

The AC'97 standard has to a great degree been replaced by the Intel 'High Definition Audio' (HDA) standard. This defines both the circuits and the PCI interface and so needs only a single set of ALSA drivers.


The Intel 'Audio CODEC '97' standard specifies audio components to be found in a mainstream PC, a standard way of connecting the components and a set of registers for controlling the components. It also defines related features such as voltages and timing characteristics.

Input can be analogue audio signals from a microphone, telephone, video, CD and other devices. Some will be mono, some stereo. Input also includes sound from the computer which is converted from digital to analogue before further processing. Upon input, each of these signals passes through an amplifier and a mute switch, both of which can be controlled by the mixer software on the computer.

Analogue output can be to stereo speakers and additional surroundsound speakers. Output of sound to the computer is provided by an analogue to digital converter. Again, all output streams can be amplified or muted, on command, from the computer. Later versions of the standard allow input and output using the Sony/Phillips Digital Interface (S/P DIF,) also frequently called IEC958 because of the international specification that subsequently standardised this format, to connect additional audio equipment.

Between input and output the standard defines two main pathways, one providing analogue output to the main stereo speakers, the other leading to the computer input. In the first, the different input streams can be combined and passed through an optional surroundsound filter, before being sent via a tone filter to the output. Digital input (PCM - Pulse Coded Modulation) can be added before or after the surroundsound step. The telephone signal is added after the surroundsound step.

For computer output the standard is somewhat different. Output to the computer is allowed from just one of the analogue inputs or from the main stereo output after the tone filter, or from a mono version of the main output. In ALSA the latter are often called the 'MIX' and 'MONO MIX' channels. Note that the digital input signal can only be captured using one of the two latter channels.

As well as the above outputs there is also a mono output channel and surroundsound channels. The mono output can be driven from the microphone input. Alternatively it can be a mono copy of the main output channel, tapped before the telephone input is added. This allows mono output to be used for telephone output without feedback developing.

The surroundsound channels can be driven by the surroundsound filter in the main pathway. Alternatively they can be driven by their own digital input stream from the computer. Additional switches are often available, for example to allow input from more than one microphone or to allow additional links between the audio channels.

High Definition Audio

This is a recent Intel specification for defining and controlling a much more sophisticated audio system. In this case the audio system is connected to a PCI or other system bus via a well specified 'HDA Controller'. This is then connected to one or more audio chips via a serial bus, the 'High Definition Audio Link'. Each audio chip (or CODEC to use Intel's terminology) can then contain a number of widgets. Typical widgets are:

On startup, each codec sends the controller information about which widgets are present and how they are connected. This allows very complex systems to be developed and controlled.

External Links

See also

Retrieved from ""

Categories: Sound cards | Documentation | Introduction | Glossary