This is what I've been working on these past few days:
Open Loudspeaker Explorer
What is it? It is an interactive, fully customizable Colaboratory notebook for the analysis, visualization, and comparison of loudspeaker measurement data that @amirm publishes along his reviews.
(This project is similar in some ways to @pierre's graph tools but the technical implementation is a bit different.)
With Loudspeaker Explorer, you can :
And now a few screenshots to get you salivating
Note that this is only the first version. I'm not done - far from it. Here's my current list of ideas:
Hope you like it
And my infinite thanks to @amirm for enabling such tools to exist by making the raw data available!
Open Loudspeaker Explorer
What is it? It is an interactive, fully customizable Colaboratory notebook for the analysis, visualization, and comparison of loudspeaker measurement data that @amirm publishes along his reviews.
(This project is similar in some ways to @pierre's graph tools but the technical implementation is a bit different.)
With Loudspeaker Explorer, you can :
- Nothing to install. Runs from your browser.
- Load any speaker that @amirm measured (with only a few exceptions due to missing data).
- You can even add speakers yourself if you have the data!
- Choose more than one speaker to compare them.
- Look at measurements in a consistent format with interactive, pretty graphs that can be zoomed in.
- Compare speakers side-by-side. This is where the tool excels:
- Frequency responses are automatically aligned (normalized) to remove sensitivity differences (this can be customized).
- Graphs shown side-by-side will always use the same scales for easy comparison (even if you zoom in!).
- The tool also generates graphs where all speakers are shown simultaneously (e.g. on-axis frequency response graph)
- Use special data processing such as on-axis normalization which emulates perfect on-axis EQ, allowing you to focus on directivity only.
- Includes the list of available speakers along with pictures, links to review, and price data.
And now a few screenshots to get you salivating
Note that this is only the first version. I'm not done - far from it. Here's my current list of ideas:
- Compute Olive preference scores, @MZKM style. This is actually the main reason why I started this project in the first place; I want to calculate the formulas from scratch to cross-check @MZKM's work (spreadsheets are really tedious to debug) and to provide more visualizations around how the score is calculated (per frequency band components, etc.) The tools I'm using (Colab, Pandas, etc.) excel at this kind of analysis.
- More interactive graphs: hide/show lines, tooltips
- Add sorting/filters settings for the speaker list (e.g. price) and improve the presentation (better use of space). Might be useful given the @amirm's impressive review cadence
- Heatmaps
- Polar maps
- Show regression lines on On-Axis and PIR charts (as well as a reference "preferred" line)
- "Toe-in" calculator: calculate Narrow Band Deviation (NBD) for each off-axis angle, automatically compute which one is the flattest, use it instead of On Axis for calculations
- "Duel" mode: when selecting two speakers, show additional charts showing the delta between the two responses
Hope you like it
Last edited: