Types of Audio Compression Techniques
The uncompressed data rate increases as more bits are used for quantization. Stereo information as opposed to mono doubles the amount of bandwidth needed to transmit a digital audio signal.
Audio Compression Techniques:
1. Pulse Code Modulation: It is the formal term for the audio digitization method that includes sampling and quantization. The resulting digitized samples can be thought of as infinitely narrow vertical ‘pulses’. PCM enables other coding techniques.
2. Differential Pulse Code Modulation: DPCM technique is based on storing the difference between consecutive samples instead of the absolute value of the sample in fewer bits. However, unlike video, the audio waveform can change rapidly and hence there is lesser temporal redundancy which means the difference signal value is also higher.
DPCM employs a predictive coding scheme followed by quantization. It computes a predicted value for a sample based on preceding samples and stores the prediction error.
3. Adaptive DPCM: It obtains further compression by dynamically varying the step size used to quantize signal differences. Large differences are quantized using large steps, and small differences using small steps, so the amount of details stored is scaled or adapted to the size of the difference. Sometimes the predictor coefficients are adaptively modified in the process known as Adaptive Predictive Coding (APC).
4. Delta Modulation: It is a simplified version of DPCM which uses only a single quantized error value. It works well with more or less constant audio signals.
5. Linear Predictive Coding: It is a typical voice coding or vocoding technique that generates a set of parameters modelling the shape and excitation of the vocal tract, not actual signals or differences. It is like MIDI where the description of the signal is sent rather than the signal itself.