master_me is an automatic audio mastering plugin for live-streaming, podcasting and internet radio stations. Introducing no latency, it takes care of master levels and ‘polishes’ the sound with a chain of effects, closely modeled after the audio chain of Klaus Scheuermann’s mastering studio in Berlin, Germany. The main concept of master_me is the combination of a leveler and a chain of dynamics processors. The leveler can be seen as a big volume knob which you would grab, when the sound is too low or too high for your taste. Signal Flow: - Pre-processing - Gate - EQ - Leveler - Knee Compressor - Multiband Mid-Side Compressor - Limiter - Brickwall Module: pre-processing The ‘pre-processing’ module contains: - a gain-slider to apply gain to the incoming signal before it hits master_me. - a ‘mono’ switch - phase switches for both left and right channels - ‘stereo-correct’ switch (custom designed process, see 'stereo-correct') Module: gate A simple noise gate with the following parameters: - threshold - attack - release Module: EQ The EQ offers some simple tools for frequency spectrum manipulation. The parameters are: - highpass frequency: the cutoff frequency of a soft highpass filter - tilt-gain: gain of a 'tilt' equalizer, which will bend the frequency spectrum either to the higher frequencies or the lower frequencies. - side-eq gain: gain parameter of the 'side-eq' which boosts a certain frequency range in the side signal in order to make the sound 'wider'. - side-eq frequency: center frequency of the side-eq - side-eq bandwidth: width of the side-eq Module: leveler The leveler is the most critical part of master_me. It will listen to the incoming signal and adjust it's volume to meet the target loudness. In order to make this as smooth and natural as possible, a complex algorithm was designed. The following parameters can be set: - target loudness: in lufs, a slider also available in 'easy' mode, next to the input meters on the left). Typical target loudness values are -18lufs for video streaming, -14lufs for podcasting, -23lufs for EBU broadcast standard. - brake threshold: this parameter is used by the leveler to detect silence. When silence is detected, the leveler 'freezes' and waits for incoming audio before it continues to adjust the loudness. - max +: determines the maximum amount of positive gain the leveler can apply to the incoming signal - max -: determines the maximum amount of negative gain the leveler can apply to the incoming signal - the brake-meter shows silence detection. Module: knee compressor The knee compressor is a slow and soft mid-side compression module. It functions as a subtile, swinging compressor. Typical equivalents in the analog domain would be a Manley Stereo VARIABLE MU® compressor or a Vertigo VSC-2. The following parameters can be set: - strength: correlating to the ratio of compression. 0% equals a ratio of 1:1, 100% equals a ratio of 1:infinity - tar-thresh: offsets the compressors threshold, dependent on the target loudness. - attack: compressor's attack time in milliseconds - release: compressor's release time in milliseconds - knee: compressor's knee in dB - link: amount of gain reduction linking between mid and side channels - ff-fb: feedforward-feedback determines, where the compressor receives it's side-chain signal from. feedforward is the input of the compressor, feedback is the output of the 'brickwall' module. - make-up: simple gain makeup after compression - dry-wet: fades between the input signal and the compressed signal - the meters show the amount of compression on each channels Module: multiband mid/side compressor Being perhaps the most complicated module of master_me, this module works like this: The audio is first converted from stereo to mid-side and then split into 8 frequency bands. The parameters apply to the lowest band (low) and the highest band (high). For the 6 bands in between, the parameters are interpolated between 'low' and 'high'. The parameters for the lowest and highest band are: - strength: correlating to the ratio of compression. 0% equals a ratio of 1:1, 100% equals a ratio of 1:infinity - tar-thresh: offsets the band's threshold, dependent on the target loudness. - attack: band's attack time in milliseconds - release: band's release time in milliseconds - knee: band's knee in dB - link: amount of gain reduction linking between mid and side channels - crossover: the lowest and highest crossover frequencies can be set here. All crossovers in between will be interpolated. - the upper row of meters shows the gain reduction for all eight mid channels - the lower row of meters shows the gain reduction for all eight side channels Module: limiter The 'limiter' it is rather a sound-shaping limiter than a clip-protection limiter. It's equivalents in the analog domain would typically be a Chandler TG-1 or a UREI 1178. Although the limiter can apply high compression ratios, it will not prevent from digital overshoots higher than threshold (which the brickwall module will take care of). The parameters are: - strength: correlating to the ratio of compression. 0% equals a ratio of 1:1, 100% equals a ratio of 1:infinity - tar-thresh: offsets the limiter's threshold, dependent on the target loudness. - attack: limiter's attack time in milliseconds - release: limiter's release time in milliseconds - knee: limiter's knee in dB - ff-fb: feedforward-feedback determines, where the limiter receives it's side-chain signal from. feedforward is the input of the limiter, feedback is the output of the limiter. - make-up: simple gain makeup after limiting. - gain reduction meter: shows gain reduction applied to the signal. Module: brickwall The 'brickwall' module is the last process in master_me's chain of modules. It is a fast brickwall limiter which will not allow any peaks above the desired 'ceiling'. The 'brickwall' process is a protection limiter and will not sound nice, if it needs to work a lot. Custom module: stereo correct This process was custom developed for the Chaos Computer Club's VOC (video operation center). It constantly checks the phase of an incoming signal. If phase == 1, all audio is made mono. If phase == -1, one channel is phase switched and the audio is monofied. This helps detect and fix broken audio signals and unwanted panning Concept and idea: Klaus Scheuermann DSP: Klaus Scheuermann (trummerschlunk), Bart Brouns (magnetophon), Robin Gareus (x42), Jakob Dübel (jkbd) GUI, Plugin: Filipe Coelho (falkTX) Logo, Artwork: Peter Schlossnikel More contributions from: Julius Smith, Dario Sanfilippo, Stéphane Letz, Romain Michon, Yann Orlarey, the Faust community.