Chromaprint

OS independent wrapper for chromaprint. There exists already a chromaprint API. However, it uses cgo and does not run on windows out of the box. This wrapper removes the cgo dependency and allows to run chromaprint by using the binary directly.

Example

package main

import (
 "fmt"

 "github.com/jo-hoe/chromaprint"
)

func main() {
 // assuming fpcalc executable (aka chromaprint) is in the
 // avaiable in $PATH. Alternativly use 
 // NewBuilder().WithPathToChromaprint("/path/to/fpcalc")
 chromaprinter, err := chromaprint.NewBuilder().Build()
 if err != nil {
  fmt.Print(err)
  return
 }
 fingerprints, err := chromaprinter.CreateFingerprints("my.mp3")
 if err != nil {
  fmt.Print(err)
  return
 }
 fmt.Printf("%+v", fingerprints)
}

Chromaprint

Chromaprint provides a fingerprint for a given audio. The aim is to provide a single identifier for a given audio input. This identifier should be robust to background noice.

Here is how it roughly works:

• audio is resampled to mono at 11025Hz
• the result is cut into individual frames
• fast fourier transform is performed to extract frequencies within each chucks
• features are extracted from the these frequencies
• these features are converted into an image
• a fingerprint is calculated from this image

Chromaprint Algorithms

Chromaprint allows to set an algorithm for the generation of fingerprint.

// where n is an integer between 1 and 5
chromaprint.NewBuilder().WithAlgorithm(n)

Each of these provides a different fingerprint for a given audio input. Each algorithm requires a minimum amount of audio data before a fingerprint can be calculated. Below is a table showing how long an audio clip has to be before the different algorithms can provide a fingerprint.

# Algorithm	Minimum seconds of audio
1	~ 3
2	~ 3
3	~ 7.5
4	~ 3.5
5	~ 2

Note, that these values are only rough approximations and differ depending on the audio input.

Dependency

It is required have a chromaprint CLI.

License

While this repository uses MIT the chromaprint license may differ (i.e. LGPL2.1).

Linting

Project used golangci-lint for linting.

Installation

https://golangci-lint.run/usage/install/

Execution

Run the linting locally by executing

golangci-lint run ./...

in the working directory