AudioFormat

Class Overview

The AudioFormat class is used to access a number of audio format and channel configuration constants. They are for instance used in AudioTrack and AudioRecord, as valid values in individual parameters of constructors like AudioTrack(int, int, int, int, int, int), where the fourth parameter is one of the AudioFormat.ENCODING_* constants. The AudioFormat constants are also used in MediaFormat to specify audio related values commonly used in media, such as for KEY_CHANNEL_MASK.

The AudioFormat.Builder class can be used to create instances of the AudioFormat format class. Refer to AudioFormat.Builder for documentation on the mechanics of the configuration and building of such instances. Here we describe the main concepts that the AudioFormat class allow you to convey in each instance, they are:

1. sample rate
2. encoding
3. channel masks

Closely associated with the AudioFormat is the notion of an audio frame, which is used throughout the documentation to represent the minimum size complete unit of audio data.

Sample rate

Expressed in Hz, the sample rate in an AudioFormat instance expresses the number of audio samples for each channel per second in the content you are playing or recording. It is not the sample rate at which content is rendered or produced. For instance a sound at a media sample rate of 8000Hz can be played on a device operating at a sample rate of 48000Hz; the sample rate conversion is automatically handled by the platform, it will not play at 6x speed.

As of API M, sample rates up to 192kHz are supported for AudioRecord and AudioTrack, with sample rate conversion performed as needed. To improve efficiency and avoid lossy conversions, it is recommended to match the sample rate for AudioRecord and AudioTrack to the endpoint device sample rate, and limit the sample rate to no more than 48kHz unless there are special device capabilities that warrant a higher rate.

Encoding

Audio encoding is used to describe the bit representation of audio data, which can be either linear PCM or compressed audio, such as AC3 or DTS.

For linear PCM, the audio encoding describes the sample size, 8 bits, 16 bits, or 32 bits, and the sample representation, integer or float.

• ENCODING_PCM_8BIT: The audio sample is a 8 bit unsigned integer in the range [0, 255], with a 128 offset for zero. This is typically stored as a Java byte in a byte array or ByteBuffer. Since the Java byte is signed, be careful with math operations and conversions as the most significant bit is inverted.
• ENCODING_PCM_16BIT: The audio sample is a 16 bit signed integer typically stored as a Java short in a short array, but when the short is stored in a ByteBuffer, it is native endian (as compared to the default Java big endian). The short has full range from [-32768, 32767], and is sometimes interpreted as fixed point Q.15 data.
• ENCODING_PCM_FLOAT: Introduced in API LOLLIPOP, this encoding specifies that the audio sample is a 32 bit IEEE single precision float. The sample can be manipulated as a Java float in a float array, though within a ByteBuffer it is stored in native endian byte order. The nominal range of ENCODING_PCM_FLOAT audio data is [-1.0, 1.0]. It is implementation dependent whether the positive maximum of 1.0 is included in the interval. Values outside of the nominal range are clamped before sending to the endpoint device. Beware that the handling of NaN is undefined; subnormals may be treated as zero; and infinities are generally clamped just like other values for AudioTrack – try to avoid infinities because they can easily generate a NaN.
  To achieve higher audio bit depth than a signed 16 bit integer short, it is recommended to use ENCODING_PCM_FLOAT for audio capture, processing, and playback. Floats are efficiently manipulated by modern CPUs, have greater precision than 24 bit signed integers, and have greater dynamic range than 32 bit signed integers. AudioRecord as of API M and AudioTrack as of API LOLLIPOP support ENCODING_PCM_FLOAT.

For compressed audio, the encoding specifies the method of compression, for example ENCODING_AC3 and ENCODING_DTS. The compressed audio data is typically stored as bytes in a byte array or ByteBuffer. When a compressed audio encoding is specified for an AudioTrack, it creates a direct (non-mixed) track for output to an endpoint (such as HDMI) capable of decoding the compressed audio. For (most) other endpoints, which are not capable of decoding such compressed audio, you will need to decode the data first, typically by creating a MediaCodec. Alternatively, one may use MediaPlayer for playback of compressed audio files or streams.

When compressed audio is sent out through a direct AudioTrack, it need not be written in exact multiples of the audio access unit; this differs from MediaCodec input buffers.

Channel mask

Channel masks are used in AudioTrack and AudioRecord to describe the samples and their arrangement in the audio frame. They are also used in the endpoint (e.g. a USB audio interface, a DAC connected to headphones) to specify allowable configurations of a particular device.
As of API M, there are two types of channel masks: channel position masks and channel index masks.

Channel position masks

Channel position masks are the original Android channel masks, and are used since API BASE. For input and output, they imply a positional nature - the location of a speaker or a microphone for recording or playback.
For a channel position mask, each allowed channel position corresponds to a bit in the channel mask. If that channel position is present in the audio frame, that bit is set, otherwise it is zero. The order of the bits (from lsb to msb) corresponds to the order of that position's sample in the audio frame.
The canonical channel position masks by channel count are as follows:

channel count	channel position mask
1	CHANNEL_OUT_MONO
2	CHANNEL_OUT_STEREO
3	CHANNEL_OUT_STEREO | CHANNEL_OUT_FRONT_CENTER
4	CHANNEL_OUT_QUAD
5	CHANNEL_OUT_QUAD | CHANNEL_OUT_FRONT_CENTER
6	CHANNEL_OUT_5POINT1
7	CHANNEL_OUT_5POINT1 | CHANNEL_OUT_BACK_CENTER
8	CHANNEL_OUT_7POINT1_SURROUND

These masks are an ORed composite of individual channel masks. For example CHANNEL_OUT_STEREO is composed of CHANNEL_OUT_FRONT_LEFT and CHANNEL_OUT_FRONT_RIGHT.

Channel index masks

Channel index masks are introduced in API M. They allow the selection of a particular channel from the source or sink endpoint by number, i.e. the first channel, the second channel, and so forth. This avoids problems with artificially assigning positions to channels of an endpoint, or figuring what the i^th position bit is within an endpoint's channel position mask etc.
Here's an example where channel index masks address this confusion: dealing with a 4 channel USB device. Using a position mask, and based on the channel count, this would be a CHANNEL_OUT_QUAD device, but really one is only interested in channel 0 through channel 3. The USB device would then have the following individual bit channel masks: CHANNEL_OUT_FRONT_LEFT, CHANNEL_OUT_FRONT_RIGHT, CHANNEL_OUT_BACK_LEFT and CHANNEL_OUT_BACK_RIGHT. But which is channel 0 and which is channel 3?
For a channel index mask, each channel number is represented as a bit in the mask, from the lsb (channel 0) upwards to the msb, numerically this bit value is 1 << channelNumber. A set bit indicates that channel is present in the audio frame, otherwise it is cleared. The order of the bits also correspond to that channel number's sample order in the audio frame.
For the previous 4 channel USB device example, the device would have a channel index mask 0xF. Suppose we wanted to select only the first and the third channels; this would correspond to a channel index mask 0x5 (the first and third bits set). If an AudioTrack uses this channel index mask, the audio frame would consist of two samples, the first sample of each frame routed to channel 0, and the second sample of each frame routed to channel 2. The canonical channel index masks by channel count are given by the formula (1 << channelCount) - 1.

Use cases

• Channel position mask for an endpoint: CHANNEL_OUT_FRONT_LEFT, CHANNEL_OUT_FRONT_CENTER, etc. for HDMI home theater purposes.
• Channel position mask for an audio stream: Creating an AudioTrack to output movie content, where 5.1 multichannel output is to be written.
• Channel index mask for an endpoint: USB devices for which input and output do not correspond to left or right speaker or microphone.
• Channel index mask for an audio stream: An AudioRecord may only want the third and fourth audio channels of the endpoint (i.e. the second channel pair), and not care the about position it corresponds to, in which case the channel index mask is 0xC. Multichannel AudioRecord sessions should use channel index masks.

Audio Frame

For linear PCM, an audio frame consists of a set of samples captured at the same time, whose count and channel association are given by the channel mask, and whose sample contents are specified by the encoding. For example, a stereo 16 bit PCM frame consists of two 16 bit linear PCM samples, with a frame size of 4 bytes. For compressed audio, an audio frame may alternately refer to an access unit of compressed data bytes that is logically grouped together for decoding and bitstream access (e.g. MediaCodec), or a single byte of compressed data (e.g. AudioTrack.getBufferSizeInFrames()), or the linear PCM frame result from decoding the compressed data (e.g.AudioTrack.getPlaybackHeadPosition()), depending on the context where audio frame is used.

Summary

Nested Classes
class	AudioFormat.Builder	Builder class for AudioFormat objects.

Constants
int	CHANNEL_CONFIGURATION_DEFAULT	This constant was deprecated in API level 5. Use CHANNEL_OUT_DEFAULT or CHANNEL_IN_DEFAULT instead.
int	CHANNEL_CONFIGURATION_INVALID	This constant was deprecated in API level 5. Use CHANNEL_INVALID instead.
int	CHANNEL_CONFIGURATION_MONO	This constant was deprecated in API level 5. Use CHANNEL_OUT_MONO or CHANNEL_IN_MONO instead.
int	CHANNEL_CONFIGURATION_STEREO	This constant was deprecated in API level 5. Use CHANNEL_OUT_STEREO or CHANNEL_IN_STEREO instead.
int	CHANNEL_INVALID	Invalid audio channel mask
int	CHANNEL_IN_BACK
int	CHANNEL_IN_BACK_PROCESSED
int	CHANNEL_IN_DEFAULT
int	CHANNEL_IN_FRONT
int	CHANNEL_IN_FRONT_PROCESSED
int	CHANNEL_IN_LEFT
int	CHANNEL_IN_LEFT_PROCESSED
int	CHANNEL_IN_MONO
int	CHANNEL_IN_PRESSURE
int	CHANNEL_IN_RIGHT
int	CHANNEL_IN_RIGHT_PROCESSED
int	CHANNEL_IN_STEREO
int	CHANNEL_IN_VOICE_DNLINK
int	CHANNEL_IN_VOICE_UPLINK
int	CHANNEL_IN_X_AXIS
int	CHANNEL_IN_Y_AXIS
int	CHANNEL_IN_Z_AXIS
int	CHANNEL_OUT_5POINT1
int	CHANNEL_OUT_7POINT1	This constant was deprecated in API level 23. Not the typical 7.1 surround configuration. Use CHANNEL_OUT_7POINT1_SURROUND instead.
int	CHANNEL_OUT_7POINT1_SURROUND
int	CHANNEL_OUT_BACK_CENTER
int	CHANNEL_OUT_BACK_LEFT
int	CHANNEL_OUT_BACK_RIGHT
int	CHANNEL_OUT_DEFAULT	Default audio channel mask
int	CHANNEL_OUT_FRONT_CENTER
int	CHANNEL_OUT_FRONT_LEFT
int	CHANNEL_OUT_FRONT_LEFT_OF_CENTER
int	CHANNEL_OUT_FRONT_RIGHT
int	CHANNEL_OUT_FRONT_RIGHT_OF_CENTER
int	CHANNEL_OUT_LOW_FREQUENCY
int	CHANNEL_OUT_MONO
int	CHANNEL_OUT_QUAD
int	CHANNEL_OUT_SIDE_LEFT
int	CHANNEL_OUT_SIDE_RIGHT
int	CHANNEL_OUT_STEREO
int	CHANNEL_OUT_SURROUND
int	ENCODING_AC3	Audio data format: AC-3 compressed
int	ENCODING_DEFAULT	Default audio data format
int	ENCODING_DTS	Audio data format: DTS compressed
int	ENCODING_DTS_HD	Audio data format: DTS HD compressed
int	ENCODING_E_AC3	Audio data format: E-AC-3 compressed
int	ENCODING_INVALID	Invalid audio data format
int	ENCODING_PCM_16BIT	Audio data format: PCM 16 bit per sample.
int	ENCODING_PCM_8BIT	Audio data format: PCM 8 bit per sample.
int	ENCODING_PCM_FLOAT	Audio data format: single-precision floating-point per sample

Public Methods
int	getChannelCount() Return the channel count.
int	getChannelIndexMask() Return the channel index mask.
int	getChannelMask() Return the channel mask.
int	getEncoding() Return the encoding.
int	getSampleRate() Return the sample rate.
String	toString() Returns a string containing a concise, human-readable description of this object.

[Expand] Inherited Methods

From class java.lang.Object Object clone() Creates and returns a copy of this Object. boolean equals(Object o) Compares this instance with the specified object and indicates if they are equal. void finalize() Invoked when the garbage collector has detected that this instance is no longer reachable. final Class<?> getClass() Returns the unique instance of Class that represents this object's class. int hashCode() Returns an integer hash code for this object. final void notify() Causes a thread which is waiting on this object's monitor (by means of calling one of the wait() methods) to be woken up. final void notifyAll() Causes all threads which are waiting on this object's monitor (by means of calling one of the wait() methods) to be woken up. String toString() Returns a string containing a concise, human-readable description of this object. final void wait() Causes the calling thread to wait until another thread calls the notify() or notifyAll() method of this object. final void wait(long millis, int nanos) Causes the calling thread to wait until another thread calls the notify() or notifyAll() method of this object or until the specified timeout expires. final void wait(long millis) Causes the calling thread to wait until another thread calls the notify() or notifyAll() method of this object or until the specified timeout expires.

   getClass().getName() + '@' + Integer.toHexString(hashCode())