Outboard Hardware Selection
We chose to model the API 212L pre-amplifier and the KT(Klark Teknik) EQP KT. Both hardware pieces are extremely popular for audio engineers and music producers, especially with treating pop-voice recordings or the acoustic guitar.
API 212L: Schematic
For the API 212L pre-amp, we found the schematics provided by API for the preamp and used LT spice to calculate the approximated transfer function for the signal processor.
EQP KT: Data Collection
For the EQP KT, we used the Audio Studios at the Duderstadt Center to take measurements of the filters in the signal processor. We used a white noise generator from Logic Pro X(Digital Audio Workstation) and measured the change in frequency response for different settings.
The following are the set values for the measurements.
Bandwidth: 0, 5, 10
High Boost Frequency: 3k, 5k, 10k, 16kHz
Low Boost Frequency: 20, 60, 80, 200Hz
High Cut Frequency: 3k, 5k, 20kHz
Low Cut Frequency: 20, 60, 80, 200Hz
All measurements were taken in intervals of 1, from 1 to 10, based on the physical label of the hardware.
EQP-KT Frequency Response Plots
Key Takeaways from EQP-KT Data Collection, Plotting, and Processing
We noticed when plotting our frequency response graphs in Matlab that they were generally pretty noisy. This is especially noticeable in Figure 1 and Figure 4. This was mitigated visually using the smoothing function in Matlab as demonstrated in Figure 2.
Additionally, the bandwidth of the high boost option is not as noticeable or responsive when compared to the low boost option as seen in Figure 4; in general, the sharper bandwidth is broader than expected when higher center frequencies are selected such as 10kHz and 16kHz, but the sharper bandwidth is more noticeable on the 3kHz and 5kHz option.
We noticed that the filter types of the low boost and high cut filters appeared to be a shelf style, which makes sense considering the use cases of such filters. Additionally, we assume that the filter type for the low cut filter is a high-pass filter. Finally, the type for the high boost is a bell filter.
During preliminary convolution testing, we also discovered that there is some unexpected high-mid frequency noise in the convolved output. This is likely related to the especially noisy high boost frequency responses, and could be mitigated by re-examining our data collection process.
Challenges
1
API's Op amp
API's audio processors are known for its unique Op-amp, which many people believe that results in the unique timbre of the processed signal. Translating the schematic to model the circuit on LT spice was definitely a challenging aspect.
2
Transparency of Audio Signal
There were some real-world constraints and risks in the data collection process for the EQP-KT since we were working with analog hardware. Also, using white noise in order to capture the frequency response introduced a subsequent amount of noise in our data.
3
Organizing Raw Data
There were a lot of measurements and audio files for the collected data, and organising the files in a orderly fashion for analysis and evaluation was also a challenge. We overcame this challenge by labeling each file with its settings and organizing the Logic project session with folders for each setting using the Bandwidth value as the main divisor.
Future Plans
For the extension after the project, we aim to finish the emulations of the two hardware and design a digital signal processor that incorporates the 212L and the EQP-KT functions into a single control.
Below is the objective checklist, where you can see the current status of our project.
Objectives Checklist
for the API 212L
for the EQP KT
✅ Model on LT spice
⏺️ Obtain Transfer function
⏺️ Create Simulation
✅ Frequency Response measurement
⏺️ Model by using filters
⏺️ Create Simulation
for the extensive
Final Outcome
⏺️ Design an effective one-button/knob-type program to apply the audio processing effect created by our two emulations.
our very own
Voice Aerator
This virtual program (VST) will be designed by integrating functions of the two pieces of hardware modeled. We aim to create a simple 'one-touch' approach for beginners in audio mixing to be able to achieve a clean vocal track without having to spend long hours changing EQ parameters or input gain levels.
We will pre-set the values of the EQ and input gain so that the VST plugin will process the audio signal so that it would sound more 'open' or 'closed' with cleaning up the unnecessary frequency bands of the signal.
The UI design and control were inspired by the "OneKnob" series digital signal processors provided by Waves.
We learned about a new VST development tool called 'JUCE'
in the process of our research into how to implement the real-life application for our emulation.