Audio Codec Interface

The codec interface is meant to be used when the codecs are secondary to a controller driver. This interface is a FIDL protocol exposed by codec drivers. In this arrangement the codec drivers are not directly exposing a streaming interface, and they are configured via the codec interface by a controller. This is a reference for driver-authors, defining the interface contract which codec drivers must implement and that controllers can use.

Notation and Terminology

In this document:

  • All indices start from 0.
  • Vectors of n elements are represented as <x0,x1,...,xn-1>, for example a vector with two elements 5 and 6 as <5,6>.
  • Vectors can be nested, i.e. <<5,6>,<7,8>> represents a vector with 2 vectors in it.
TermDefinition
CodecA real or virtual device that encodes/decodes a signal from
: : digital/analog to/from analog/digital including all :
: : combinations, e.g. digital to digital. Example codecs include :
: : DAC-Amplifiers combos and ADC converters. :
ControllerThe part of a system that manages the audio signals, for
: : example an SOC's audio subsystem or an independent sound card. :
DAIDigital Audio Interface. Interface between audio HW, for :
: : instance a TDM or PDM link between controllers and codecs. :
Frame SyncA DAI signal that marks frame boundaries, a.k.a. LRCLK, SYNC.
SclkA DAI signal used to mark the data line(s) bits transferring, :
: : a.k.a. SCK, BCLK. :
MclkMaster clock, a DAI signal sometimes needed to provide a clock
: : to codecs. Sometimes Sclk is used as the Mclk (or Mclk is :
: : derived from the Sclk within the codec). :
FrameThe representation of a single moment in time across data, :
: : frame sync and sclk in the DAI. :
Frame formatA frame's data, frame sync and sclk arrangement, e.g. location
: : of the frame sync w.r.t. samples in the data line(s). :
SlotWithin a frame, the bits reserved for a sample. A slot may be
: : bigger than needed to hold the samples, e.g. 32 bits slot :
: : holding 24 or 16 bits samples. :
ChannelA single source or destination of audio samples, usually
: : to be rendered by a single speaker or captured by a single :
: : microphone. Within a DAI every frame will contain samples in :
: : a fixed number of slots for the same fixed number of channels. :
SampleA digital representation of sound taken at a particular time.

Basic Operation

The functionality provided by the codecs is divided into:

The controller is responsible for configuring and controlling the codecs. Codecs advertize capabilities and a controller determines how they are used as described below. The controller can control the codec's state, such as through the reset function. A reset is required to get codecs to an initialized state. Note that the codec drivers are expected to perform their own shutdown, just like any other driver.

Codecs are composite devices that provide the codec protocol to controllers. It is expected that only one controller uses a codec's protocol, and one controller may use multiple codecs at once.

The simple-codec library facilitates writing and using simple codec drivers implementing the codec protocol.

Protocol definition

The codec protocol is defined in FIDL at codec.fidl.

Note that because the DDK does not currently provide a way to directly get a FIDL channel for communication, we define a way to get a channel via Banjo at ddk.protocol.audio.

Many codec protocol operations are “fire-and-forget”, i.e. they do not expect a reply. Codec protocol operations with a reply are not considered completed until the reply of the function is received, and not considered completed successfully unless the reply contains a status ZX_OK.

Main Controls

A codec can be reset by a controller at any time by issuing the Reset function.

The GetInfo function retrieves information from the codec including:

  1. A unique and persistent identifier for the codec unit, e.g. a serial number or connection path.
  2. The manufacturer name.
  3. The product name.

The codec operation can be started and stopped at any time with the Start and Stop functions. By default the codec state is stopped, so ‘Start’ must be issued at least once for the codec to be fully operational. Stopping the codec operation can be used for example to change the DAI configuration safely avoiding glitches or errors in the codec operation.

Bridged Mode

Before specifying the DAI format the controller must query the codec for its bridging capabilites. If the codec is bridgeable, then the controller must enable or disable bridging based on its knowledge of the system configuration. Note that this is a singular property of a codec, i.e. a codec either supports bridging or not, and it can be set in bridged mode or not. This protocol allows configuring as bridged only 2 channel stereo codecs, with the 2 outputs of the codec electrically bridged.

DAI Format

The DAI Format related protocol functions allow the codec to list its supported formats for the DAI. The supported formats may include multiple sample formats, rates, etc. Each codec advertises what it can support and the controller mandates what DAI Format is to be used for each codec.

To find out what formats are supported by a given codec, the controller uses the GetDaiFormats function. The codec replies with a vector of DaiSupportedFormats, where each DaiSupportedFormats includes:

  1. A vector of number of channels. This lists the number of channels supported by the codec, for example <2,4,6,8>. A stereo codec reports a vector with one element <2>. Note that a codec that takes one channel and outputs its contents in all its outputs (e.g. 2 for a stereo amplifier) would report a vector with one element <1>, if it supports either one or two input channels, it would report a vector with two elements <1,2>.
  2. A vector of sample formats. DAI sample formats, e.g. PCM_SIGNED.
  3. A vector of frame formats, for example STEREO_LEFT and STEREO_RIGHT.
  4. A vector of rates. Frame rates, for example 44100, 48000, and 96000.
  5. A number of bits per slot. Number of bits in each slot in the DAI, e.g. 32 bits per slot.
  6. A vector of bits per sample. Sample widths, e.g. 24 bits per sample.

When not all combinations supported by the codec can be described with one DaiSupportedFormats, the codec returns more than one DaiSupportedFormats in the returned vector.

For example, if one DaiSupportedFormats allows for 32 bits samples at 48KHz, and 16 bits samples at 96KHz, but not 32 bits samples at 96KHz, then the codec will reply with 2 DaiSupportedFormats: <<32bits>,<48KHz>> and <<16bits>,<96KHz>>. For simplicity, this example ignores parameters other than rate and bits per sample. In the case where the codec supports either 16 or 32 bits samples at either 48 or 96KHz, the codec would reply with 1 DaiSupportedFormats: <<16bits,32bits>,<48KHz,96KHz>>.

Additionally, it is assumed that bits per sample is always smaller or equal to bits per slot. Hence, a codec can report <<16bits_per_slot,32bits_per_slot>,<16bits_per_sample,32bits_per_sample>> and this does not imply that it is reporting that 32 bits per sample on 16 bits samples is valid, it specifies only the 3 valid combinations:

  1. 16 bits slot with 16 bits samples
  2. 32 bits slot with 32 bits samples
  3. 32 bits slot with 16 bits samples

Using the information provided by the codec in IsBridgeable and GetDaiFormat, what is supported by the controller, and any other requirements, the controller specifies the format to use in the DAI with the SetDaiFormat function. This functions takes a parameter that specifies:

  1. A number of channels. This is the number of channels to be used in the DAI (for instance number of channels on a TDM bus, i.e. “on the wire”). For I2S this must be 2.
  2. A vector of channels to use. These are the channels in the DAI to be used by the codec. For I2S this must be a vector with 2 indexes <0,1>, i.e. both left and right channels are used. In bridged mode this will list only the one channel to be used by the codec, for example a codec’s stereo amplifier output bridged into one electrical mono output from the right channel of an I2S DAI would list only channel <1>. If not bridged, a codec with multiple electrical outputs that is configured with one channel in SetDaiFormat is expected to replicate the samples in this mono input on all its outputs.
  3. A sample format.
  4. A frame format.
  5. A frame rate.
  6. A number of bits per slot.
  7. A number of bits per sample.

Once SetDaiFormat is successful, the DAI format configuration is considered completed and samples can be sent across the DAI.

TODO(andresoportus): Add DAI format loss notification support once asynchronous notifications are added to Banjo.

Gain Control

Gain related support by any given codec is returned by the codec in response to a GetGainFormat function in the GainFormat structure. The controller can control gain, mute and AGC states in a codec using the SetGainState function.

Clients may request that codecs send them asynchronous notifications of gain state changes by using the WatchGainState command. The driver will reply to the first |WatchGainState| sent by the client and will not respond to subsequent client |WatchGainState| calls until the gain state changes from what was most recently reported.

Plug Detect

Clients may request that codecs send them asynchronous notifications of plug state changes by using the WatchPlugState command if the CAN_ASYNC_NOTIFY flag was sent by the driver in GetPlugDetectCapabilites. I.e. drivers for codecs which do not set the CAN_ASYNC_NOTIFY flag are free to ignore the WatchPlugState sent by clients. Drivers with CAN_ASYNC_NOTIFY set will reply to the first |WatchPlugState| sent by the client and will not respond to subsequent client |WatchPlugState| calls until the plug state changes from what was most recently reported.

Power Control

TODO(fxbug.dev/63522).

Peripheral Control

TODO(fxbug.dev/64878).

Signal Processing Control

TODO(fxbug.dev/64877).

Content Protection

TODO(fxbug.dev/63523).