AN295
0x00,
// bDelay (none)
bDelay
tells the host whether the audio stream data should be delayed to synchronize with some other audio
stream. The value is in units of USB frames. The Tone Generator requires no delay, as it is not synchronized with
any other data stream.
0x0100
// wFormatTag (PCM format)
The host uses the wFormatTag to determine how to decode the incoming audio stream. The document “USB
Audio Data formats” contains tag definitions.
4.1.11. Format Type Audio Descriptor
The next Descriptor further defines the formatting of the audio stream. The document Universal Serial Bus Class
Definition for Audio Formats defines fields in this Descriptor. The firmware defines this Format Type Descriptor as
follows:
0x0B,
// bLength (11)
0x24,
// bDescriptorType (CS_INTERFACE)
0x02,
// bDescriptorSubtype (FORMAT_TYPE)
0x01,
// bFormatType (TYPE_I)
0x02,
// bNrChannels (2)
0x02,
// bSubFrameSize (2)
// The next field should be 10, but 16 works with more standard software
0x10,
// bBitResolution (16)
0x01,
// bSamFreqType (1 sampling frequency)
0x80,
// 48,000 Hz (byte 0)
0xBB,
// 48,000 Hz (byte 1)
0x00
// 48,000 Hz (byte 2)
0x01,
// bFormatType (TYPE_I)
The document “Universal Serial Bus Class Definition for Audio Formats” divides audio streams into three types.
Type I covers streams where audio is constructed on a sample-by-sample conversion scheme, such as PCM. Type
II covers streams that do not preserve a sample-by-sample conversion scheme, but instead compress multiple
audio samples onto a number of bits that can then be reconstructed to an approximation of the source audio. Type
III includes audio streams that share some properties of Type I and properties of Type II, such as when non PCM
streams are packetized into a single “pseudo-stereo” sample.
0x02,
0x02,
0x10,
0x01,
0x00,
0x77,
0x01
// bNrChannels (2)
// bSubFrameSize (2)
// bBitResolution (16)
// bSamFreqType (1 sampling frequency)
// 96,000 Hz (byte 0)
// 96,000 Hz (byte 1)
// 96,000 Hz (byte 2)
This field tells the host that the samples represent a left and a right signal.
This field shows that each audio sample will consist of two bytes, one for the left channel and one for the right.
This value tells the host that each audio sample will consist of 16 bits of data.
This field tells the host that the audio stream will sample at only one frequency.
These three bytes give the host the sampling frequency of the audio stream in units of Hz.
0x00BB80
equals
48,000 in hexadecimal.
Note:
The sampling rate is set to 24 kHz, but the use of two audio channels effectively doubles the sampling rate.
12
Rev. 0.1
SILABS [ SILICON LABORATORIES ]
