Automatic mixers (Automixers) balance multiple sound sources based on each source’s level, quickly and dramatically adjusting the various signal levels automatically. They lower the hiss, reverberation and other extraneous noise that occur when several microphones operate simultaneously. Typically Automixers should only be used for installations with multiple microphones, and they offer the following benefits:

• Improved Signal to Noise Ratio (SNR) by limiting the Number of Open Microphones (NOM)
• Improved gain before feedback by providing gain attenuation with respect to the NOM (i.e. NOM attenuation)
• Reduced microphone comb filtering by limiting voice pickup to a single microphone

Without an Automixer	With an Automixer
Poor SNR with all microphones picking up ambient noise Poor gain before feedback with multiple potential feedback loops The voice signal is picked up by multiple microphones and suffers from multi-path interferences (also called comb filtering)	Optimized SNR by attenuating unused microphones (dotted lines), hence lowering noise pickup Optimized Gain before feedback with the introduction of NOM attenuation Reduced comb filtering with single microphone pick up

Automixer block diagram

1. First, the microphone input signal is split into a side chain filter for RMS detection and threshold comparison.
2. The RMS level of the signal is sent to a Threshold Calculator which computes the current threshold for all channels. The new threshold value is then updated back to the side chain detector for each channel of the Automixer. In other words, the exact same threshold is applied to input channels of all microphones.
3. Using the updated threshold, the side chain controls the VCA by deciding the microphone’s state.
   • If the RMS level is above threshold, the Mic state is ‘Open’ (the VCA applies no attenuation to the signal)
   • If the RMS level is below threshold, the Mic state is ‘Attenuated’ (the VCA applies an ‘Off Attenuation’ to the signal)

   Note: Biamp’s Automixer algorithm attenuates the signal using a ramped VCA. The resulting behavior is a smooth transition from one state to the other.

4. The microphone’s state (Open/Attenuated) may also drive a corresponding logic output. (Ch1 starting from the left side)
5. The threshold decision in side chain sends the microphone state to a NOM Calculator, which simply computes a global NOM count for the entire Automixer. This ‘Current NOM’ is then used to calculate the NOM Attenuation.
6. Finally, all channels are mixed to a common Mix bus (labeled “M”) on the output side.
7. One-to-one direct outputs (not mixed with others) are also available at two different stages of the audio chain.

Adaptive Threshold Sensing (ATS)
The ATS method refers to the decision making process of the microphone’s state. The ability to discriminate the ambient noise from the beneficial signal (i.e. voice of speaker) is based on constant scanning of the input channels for the highest RMS level. The algorithm is said to be “adaptive” as it dynamically adjusts its threshold depending on the overall ambient noise floor. This highlights how much proper gain structure matters to the performance of the algorithm. The usage of levelers is strongly recommended on the input side to provide consistent signal to the Automixer.

NOM Attenuation
A simple formula determines the NOM Attenuation: “Attenuate by 3dB for every doubling of the Number of Open Microphones”, which translates to the following:

Here are some examples to better illustrate the NOM attenuation:

• 1 active microphone means no attenuation: 10log(1) = 0dB
• 2 active microphones translates to doubling the NOM: 10log(2) = 3dB of attenuation

Once the limit of 8 active microphones is reached (9dB attenuation), the NOM count and attenuation maximum is reached and will not increase further (hopefully the eight people trying to talk at once will have figured that out by then…).

More information about automixers, including advanced programming tips, can be found in our archived newsletter. The companion .dap file is also available for download. Happy programming!