Audio Effects and Signal Processing PDF

3 - Audio Effects and Signal Processing Sound Design - Games and Multimedia ©2024 Miguel Negrão CC BY-NC-ND 4.0 Outline Audio Filters Dynamic Range Processing Delay, Reverb and Convolution Miscelanious Modulation effects Distortion Time/Pitch shifting Simulation Audi...

3 - Audio Effects and Signal Processing Sound Design - Games and Multimedia ©2024 Miguel Negrão CC BY-NC-ND 4.0 Outline Audio Filters Dynamic Range Processing Delay, Reverb and Convolution Miscelanious Modulation effects Distortion Time/Pitch shifting Simulation Audio Filters (EQ) An audio filter shapes the spectrum of a sound A filter can boost, attenuate or remove certain frequencies.  A filter is characterized by its frequency response. The frequency response consists of the amplitude response and phase response. Because an ideal filter is a linear time- invariant system (LTI) if a sine wave is sent as input you get a sine wave at output, perhaps with different amplitude and phase. LTI system With an LTI system we have.  track 1 EQ track 2 EQ track 1 EQ track 2 Reverb and Eq are LTI systems, therefore they can be used in a bus, applied to multiple tracks (saving CPU).  Audio Filter Conceptually: By measuring the amplitude for sine waves with different frequencies you get the amplitude response. LTI system  Audio Filter Conceptually: By measuring the phase (delay) for sine waves with different frequencies you get the phase response. Amplitude response is usually the only one that is taken into account in sound engineering. Audio Filters Types  Low-pass filter 10 Cutoff frequency 0 −3.01 dB −10 Slope: −20 dB/decade −20 Gain (dB) −30 −40 −50 Passband Stopband −60 0.001 0.01 0.1 1 10 100 1000 Angular frequency (rad/s) Cutoff frequency Rate of frequency roll-off (slope, order) - 6, 12, 18, 24 dB/octave  High-Pass Filter 10 Cutoff frequency 0 −10 Slope: −20 dB/decade −20 Gain (dB) −30 −40 −50 −60 0.001 0.01 0.1 1 10 100 1000 Angular frequency (rad/s) Cutoff frequency Rate of frequency roll-off (slope, order) - 6, 12, 18, 24 dB/octave  Parametric equalizer +12 Center frequency +6 Amplitude (boost or cut - dB) 0 Quality factor (Q) Q bandwidth (inverse -6 relationship) -12 20 100 1k 10k 20k Bigger Q smaller bandwidth narrower filter  Band-pass filter Band-pass filter 0.35 90 0.3 60 0.25 30 Uin/Uout 0.2 Δφ(°) 0 0.15 -30 0.1 0.05 -60 0 -90 1 10 100 1,000 10,000 100,000 1e+06 f(Hz) abs(H(j ω )) arg{H(j ω )} Center frequency bandwidth = high cutoff - low cutoff Sharpness or quality factor (Q)  Band-reject or Notch filter 1.0 0.8 |H(jω)| 0.6 0.4 0.2 0.0 90 ∠H(jω) / deg 45 0 −45 −90 −1 0 1 2 3 10 10 10 10 10 −1 ω / rad s Center frequency bandwidth = high cutoff - low cutoff Sharpness or quality factor (Q)  shelving - lo-shelf, hi-shelf +12 +6 0 -6 -12 20 100 1k 10k 20k Center frequency Amplitude (boost or cut - dB) Order (slope) - 6, 12 dB/octave Dynamic Range Processing Dynamic range processing changes the envelope of signals.  Compression or Dynamic Range Compression Compression reduces the difference between loudest and softest sound. Compression parameters: Threshold Ratio Ratio 2:1 Threshold Compression +9 Input level parameters: +6 Output level Threshold +3 0 Threshold -3 Level Time Attack Release Ratio Phase Phase Compression parameters: Attack Release Input level +9 +6 Output level Threshold +3 0 -3 Level Time Attack Release Phase Phase  Compression parameters: Makeup gain Compression always reduces the level. Overall the sound will be perceived as "less loud". This can be compensated by increasing the gain by a constant factor (the makeup gain). Original Compressed Make up gain  Dynamic processor graph Threshold 1:1 Gain Reduction 2:1 Output Output Level Level (dB) (dB) 4:1 ∞:1 Threshold Input Level (dB) Input Level (dB) Compression parameters: Knee Hard Knee Output Level Soft Knee (dB) Threshold Input Level (dB)  Implementation Audio In Audio Out Amp Side Gain reduction Chain measure Feed-forward design  Multi-band compressor Splits audio using a set of overlapping filters. Each frequency band has a different compressor. Result is summed.  Compression and LTI Compression is not a linear time-invariant system. Summing and compressing is not equivalent to compressing then summing. Compression followed by EQ is not equivalent to EQ followed by compression. Limiting  A limiter is a compressor with a very high ratio. Output Level (dB) It is used to avoid clipping. Threshold Input Level (dB) Original Signal Distortion Threshold Hard Clipping (Limiting with zero attack and release) Limiter with zero attack and moderate release (brickwall) Limiter with moderate attack and release Soft Clipping Look-ahead limiter Noise Gate A noise gate is used to clean up signals which have constant background noise.  Noise Gate Input Level Threshold Range Level Output Level Time Attack Hold Release Noise gate parameters: Threshold Attack Release  Expander Like a noise gate but with a controllable ratio. Increases dynamic range Makes quiet sounds quieter. Compressor Noise Gate Expander output output output input input input  De-esser Use case: reduce the prominence of sibilant consonants such as "s", "z", "ch", etc.  Delay Sum with original signal Delay  Delay Sum with original signal Small delay comb filtering Large delay echo  2.5 α = −1 α = −0.75 α = −0.5 2 Comb filtering 1.5 ⎜H( ejω)⎜ Frequency 1 response 0.5 0 0 π 2π 3π 4π 5π 6π 7π 8π 9π 10π ω (rad/s) Reverb The reverb effect simulates a sound source inside a room. Direct Sound dB Early reflections Reverb tail time youtbe animation  Reverb time is the time it takes dB Direct Sound Early reflections for the signal to drop by 60dB Reverb tail after the sound source is time silenced. Some parameters of reverb effects: Reverb Time (s) Room size Reverb Type (room, hall, plate, etc.) Pre-delay (initial delay) Damping Low cut High cut  Implementation Analogue: spring, plate, chamber. Digital Synthetic (using delay networks, physical modeling, etc) Convolution (use impulse response of room)  Convolution Play an impulse in a room and record the result. Using the impulse response the reverb of the room can be applied to any sound using convolution. https://www.youtube.com/watch?v=cQYpZQXiLNo Modulation effects  Flanger Implemented by modulating delay time with a low frequency oscilator. Comb filter where the nulls sweep across the spectrum. Delay https://www.youtube.com/watch?v=Ici_YOVDl_0  Chorus Chorus is when several versions of the same sound with variations are played. This is what happens when several people sing the same song together. This effect can be simulated with digital processing.  Phaser Sum original signal with signal from a series of all-pass filters. This also creates notches like a flanger but they are not equally distributed.  tremolo Amplitude modulation with LFO.  Distortion Amplitude distortion happens when the output amplitude is not a linear function of the input amplitude.  Example: clipping. Original Signal Distortion Threshold Hard Clipping (Limiting with zero attack and release) Limiter with zero attack and moderate release (brickwall) Limiter with moderate attack and release Soft Clipping  Amplitude distortion can be manifested in Harmonic distortion: the creation of harmonics ( , , , etc) of the fundamental frequency of a sine wave input to a system. Usually measured as Total Harmonic Distortion (THD) which is a ratio or percentage. Intermodulation distortion: input sine waves with frequencies and output additional sine waves with frequencies , , , , etc.  Time/Pitch shifting Changing play rate will change pitch. There are techniques for changing pitch and time independently (time stretching and pitch shifting), but they always introduce some artifacts.  Simulation Guitar/Bass amplifier and cabinet Analog distortion pedals Analog delays. Vinyl  Attribution: Images taken from wikipedia with Creative Commons Attribution- Share Alike 3.0 Unported licence.  Recursos para estudo: Digital Sound and Music - Chapter 7 - Audio Processing (free, pdf moodle)  Recursos para estudo (extra): Português: "Introdução à Engenharia de Som"; Nuno Fonseca; FCA; 2012 - Capítulo 4 - Efeitos English: "Modern Recording Techniques"; David Miles Huber, Robert E. Runstein; Focal Press; 8th edition; 2013 - Chapter 15 - Signal Processing