Pulse Analysis

Using an audio file, measure pulses for timing deviation

This would generally be used to measure the timing accuracy of drum machines, sequencers and other music electronics. Inspired by the Inner Clock Systems Litmus Test

The package libsndfile must be installed first. It's available in Homebrew, APT, Yum as well as many other package managers. For those who wish to compile themselves or need more information, follow the link above for more information

Install the gem itself using

gem install pulse-analysis

Or if you're using Bundler, add this to your Gemfile

gem "pulse-analysis"

Pulse-Analysis operates on a single input audio file at a time. This file must be in an uncompressed format such as WAV or AIFF.

In keeping with conventions established by the Litmus Test, audio input files must be at least 48k sample rate.

Mono audio files are recommended. If a stereo file is used, only the left channel will be analyzed.

It's recommended that the audio file have a pulse rate of 16th notes at 120 BPM and be around 10 minutes long. In other words, to produce the best results, use only a single repetitive pulse-like sound (eg snare drum) striking 16th notes.

Example audio files are included in the repository and can be found here

pulse-analysis /path/to/a/sound/file.wav

This will run the program and output something like

[/] Reading file /path/to/a/sound/file.wav Done!
[/] Running analysis Done!
[\] Generating Report Done!

+------------------------+-------------------------+-------------+
|                         Pulse Analysis                         |
+------------------------+-------------------------+-------------+
| Item                   | Value                                 |
+------------------------+-------------------------+-------------+
| Sample rate            | 88200 (Hertz)                         |
| Length                 | 4310 (Number of pulses) | 9m0s (Time) |
| Tempo                  | 119.7546 (BPM)                        |
| Longest period length  | 11326 (Samples)         | 128.41 (ms) |
| Shortest period length | 10747 (Samples)         | 121.85 (ms) |
| Average period length  | 11047.5949 (Samples)    | 125.26 (ms) |
| Largest abberation     | 544 (Samples)           | 6.17 (ms)   |
| Average abberation     | 160.2981 (Samples)      | 1.82 (ms)   |
+------------------------+-------------------------+-------------+

2.4.0 :002 > require "pulse-analysis"
 => true

2.4.0 :003 > PulseAnalysis.report("/path/to/a/sound/file.wav")
 => {
   :file=>{
     :path=>"/path/to/a/sound/file.wav"},
     :analysis=>[
       {
         :key=>:sample_rate,
         :description=>"Sample rate",
         :value=>{
           :unit=>"Hertz",
           :value=>88200
         }
       },
       {
         :key=>:tempo,
         :description=>"Tempo",
         :value=>{
           :unit=>"BPM",
           :value=>119.9371
         }
       },
       ...

With the subjective nature of the data at hand, this program is not meant to reflect poorly on any musicians, companies, hobbyists or anyone whose product has measurable timing. After all, variation in timing may be desirable in musical context.

Additionally, it's recommended that results be independently verified. The configuration of a particular device or recording environment can cause variation in timing accuracy.

Licensed under Apache 2.0, See the file LICENSE

Copyright (c) 2017 Ari Russo