Audio Streams and Buffering

• In computing, a stream refers to a sequential flow of data moving from one point to another, such as from RAM to a hard drive.

Audio Streams

• Audio devices use a quartz crystal to produce a regular, timed audio callback, which is a special function that runs on a regular basis.
• Audio streams are divided into chunks of samples or buffers for processing.

Buffering

• Buffering is used because modern operating systems and software have heavy computational demands
• Buffering is used to ensure continuous audio processing, even if the processor is temporarily unavailable.
• Audio signal data is passed in small buffers or blocks of samples.
• A DAW determines the number of audio samples in each block for a plugin to use, typically ranging from 32 to 1024 samples or more.
• Smaller buffers result in lower latency, which is the delay from input to output, a critical factor for audio effects where some can detect latencies as low as 2ms.
• The goal is to use the smallest buffer possible before dropouts occur, where the buffer is emptied faster than it can be refilled.

Plugin Efficiency

• Efficient code is crucial because a DAW at 44.1kHz with a buffer size of 256 processes one block about every 5.8ms, requiring algorithms to run within that time.
• Sample rate and buffer size are related; increasing the sample rate with a fixed buffer size decreases the effective time length of each buffer and reduces latency.
• A 256 sample buffer at 44.1kHz represents about 5.8ms of audio, while at 88.2kHz, it represents about 2.9ms, significantly increasing CPU load.
• Increasing buffer size with a fixed sample rate increases latency but lowers CPU load.
• This relationship between sample rate, buffer size, latency, and CPU load directly impacts processing load.

Audio Buffers

• Audio buffering uses a First In, First Out (FIFO) scheme, meaning the first block of audio pushed into an input buffer is the first one out.
• FIFO input buffers receive data from the input device and feed the data to the program, while output FIFO buffers send data to the output device.

JUCE AudioBuffer Class

• The AudioBuffer class in the JUCE library encapsulates a multichannel audio buffer and is passed as a parameter reference variable of processblock().
• The AudioBuffer class is a template class, which specifies the type of data the buffer holds.

Reference Parameters

• Call by value involves passing a copy of the data to the function, so changes do not affect the original.
• Call by reference allows the function to directly access the original data, enabling efficient manipulation of the original block of samples.
• The AudioBuffer must accommodate the maximum number of input or output channels used by the audio processor.
• Initially, the buffer is populated with incoming audio data, which is then replaced by processed data for output.
• The DAW passes the AudioBuffer to the processor, which eliminates the need to manage channels or memory in most cases.
• Additional buffers may be needed for temporary internal processing or other plugin features.
• DAWs allocate and manage memory and it allows users to simply process the data within the provided buffer.
• The number of input and output channels in the buffer is detected by the DAW.
• If there are more output than input channels, only the output channels corresponding to an input channel contain valid data; extra output channels typically contain undefined or silent data.

JUCE Audio Plug-in Components

• ProcessBlock() is a special function in JUCE that is called regularly by the DAW whenever it has new audio to process.
• ProcessBlock() is where most processing code is placed.
• ProcessBlock() needs to run efficiently.

Callback Function - processBlock

• Each channel can be accessed individually using the channel index, where the left channel is index 0, and the right channel is index 2.

Panning Theory

• Linear amplitude panning is a simple method for amplitude panning, producing a linear crossfade between left and right channels.
• Linear power panning can cause sound levels to dip when a source is positioned in the center due to perceived loudness being tied to the intensity of sound waves.
• Constant power panning maintains a constant level of sound as a sound source is panned left to right.