Components of the Audio CMU

After additional reading in the past two weeks, I’ve further refined the design of the audio CMU to include biologically accurate functionalities of the pinna and ear canal. I was also focused on the coarse implementations of all audio CMU subcomponents.

Development Activity - https://github.com/akhil-reddy/beads/graphs/commit-activity
Building the outer ear components - https://github.com/akhil-reddy/beads/blob/main/beads/core/cmu/sequencing/receive/audio.py
Building the OHC components - https://github.com/akhil-reddy/beads/blob/main/beads/core/cmu/sequencing/combine/audio.py
Building the IHC components - https://github.com/akhil-reddy/beads/blob/main/beads/core/cmu/sequencing/transforms/audio.py
Building the auditory nerve components - https://github.com/akhil-reddy/beads/blob/main/beads/core/cmu/transportation/audio.py

Please note that some code (class templates, function comments, etc) is AI generated, so that I spend more of my productive time thinking and designing. However, I cross-verify each block of generated code with its corresponding design choice before moving ahead.

Algorithms / Development

Push implementation and transportation

Pinna implementation v2

  1. Given (biologically accurate) notch frequencies / notch depths, employ 2nd order Butterworth bandstop signal design for smooth filtering
    • Transform coefficients to include compression
  2. Given (biologically accurate) peak frequencies / peak gains, employ 2nd order Butterworth bandpass signal design for smooth filtering
    • Transform coefficients to include gains
  3. Sequentially apply all notches and peaks to the incoming signal
Ear canal implementation v2

  1. Given (biologically accurate) measurements of the human ear canal, generate its appropriate bandwidth and frequency
  2. Employ 2nd order Butterworth bandpass signal design for smooth filtering
    • Transform coefficients to include gains
  3. Boost frequencies around canal frequency by approx. gain decibels
OHC amplifier implementation v1

  1. For each channel, sharpen using a Dual‑Resonance Nonlinear (DRNL) model having linear and nonlinear pathways
IHC implementation v1

  1. Half wave rectification of the input signal
  2. Lowpass filtering of signal
Ribbon synapse implementation v1

  1. Power law adaptation of the IHC output
Spike train implementation v1

  1. Convert the adapted signal from ribbon synapse into a spike train using leaky integrators

Next Steps

Push implementation and transportation

  1. Implementing the inner ear CMU components

Deployment

  1. Overlaying video frames onto the retina, including code optimization for channel processing
  2. Overlaying audio clips onto the cochlea, including optimization for wave segment processing
  3. Parallelization / streaming of cellular events via Flink or equivalent

Building the Environmental Response System (ERS)

  1. Building the visual cortex
  2. Building the auditory cortex
  3. Neurotransmitters - Fed by vision’s bipolar and amacrine cells, for example, to act on contrasting and/or temporal stimulus. Neurotransmitters can be added later during ERU development

Audio CMU Components

Component Functionality
Pinna Given its asymmetric structure, the pinna (external ear cartilage) helps with spatial hearing by applying different decibel boosts (through natural resonance) to sounds coming from different directions
Ear canal The ear canal’s natural frequency helps boost mid-range frequencies and consequently aids in consonant / speech recognition
Outer hair cells They sharpen / tune frequencies so that subtle pitch differences (in speech and music, for example) are noticeable. Additionally, they boost low frequency sounds while compressing high frequency ones, to improve cochlear sensitivity
Inner hair cells The core transducer that converts sound waves (which have passed through different media) to electric signals
IHC neural transducer Ion channels that react to sound waves and stimulate spike trains
Ribbon synapse Aids in rapid release of neurotransmitters / stimulus to the auditory nerve
Efferent modulator Feedback mechanism from the brain (through neurotransmitters) to modulate the OHC gain (by adjusting elasticity of OHC cells on the basilar membrane)
Spiral ganglion neurons Analogous to the retina’s ganglion cells; a bundle of them form the auditory nerve that ends up in the auditory cortex

Audio CMU Design Updates

  1. Outer ear - After a review of related biology research, I realized that the outer ear has functionalities that aid in sound localization and resonant amplification (of consonants). These should help the brain with much needed context for richer memory as well as other higher cognitive functions. In line with the goal of this project, I aim to implement these functionalities as accurately as possible
  2. Middle ear - A digital microphone already transforms sound energy into electrical signals, so middle ear functionalities aren’t the most critical to implement. Moreover, the middle ear’s contribution to balance and acoustic protection falls outside the scope of the project
Created Apr 27, 2025