App idea: Ranking the best headphones for raw frequency response with AutoEq data

Jaakko Pasanen ranked headphones. You might be able to rank 3,937 of them with your target.

1. Find/make your favorite target.
2. For each headphone in AutoEq, calculate deviation from the target.
3. Order headphones by rank.

Jaakko Pasanen's AutoEq data has 3,937 headphones, and a bunch of targets. The headphone and target data would need to be organized and in the same format. I think it is.

Example data
target

• CSV
• graph

headphone CSV

ideas for getting the target

1. Use one of the AutoEq targets.
2. Draw a target graph. Use Jaakko's instructions to get a CSV file from the graph.

If I have time, I'll make an app to do this.

 

Hi David, you should read @purr1n 's recent comments in the HD8XX thread. He gave some thought to which and what kinds of deviations matter more.

I'd be interested in a ranking after optimized EQ.

 

Do you have any formula ideas for the deviation for a headphone?
All the CSVs have a "frequency" and "raw" column, the same number of rows, and row numbers that match the same frequencies.
my idea:
Variables indicate the type of CSV, row number, and column name.
target[i,raw] means the target CSV, row i, column raw
deviation for a row = the absolute value of target[i,raw] - headphone[i,raw]
Sum the deviation for all rows.

 

Sum of absolute deviations is not ideal.

Multiple sharp deviations at different locations are much worse than a wide deviation at one end, even though the sum of deviations would be the same. Peaks are much worse than troughs. Double sized peaks are more than 2x worse...