mdlayher/waveform

Go package capable of generating waveform images from audio streams. MIT Licensed.

waveform

Go package capable of generating waveform images from audio streams. MIT Licensed.

This library supports any audio streams which the azul3d/engine/audio package is able to decode. At the time of writing, this includes:

• WAV
• FLAC

An example binary called waveform is provided which show's the library's usage. Please see cmd/waveform/README.md for details.

Examples

Here are several example images generated using waveform. Enjoy!

Generate a waveform image, and scale it both vertically and horizontally.

$ cat ~/Music/02\ -\ Peace\ Of\ Mind.flac | waveform -x 5 -y 2 > ~/waveform.png

Apply a foreground and background color, to make things more interesting.

cat ~/Music/02\ -\ Peace\ Of\ Mind.flac | waveform -fg=#FF3300 -bg=#0099CC -x 5 -y 2 > ~/waveform_color.png

Apply an alternate foreground color, draw using a stripe pattern.

cat ~/Music/02\ -\ Peace\ Of\ Mind.flac | waveform -fg=#FF3300 -bg=#0099CC -alt=#FF9933 -fn stripe -x 5 -y 2 > ~/waveform_stripe.png

Apply an alternate foreground color, draw using a random fuzz pattern.

cat ~/Music/02\ -\ Peace\ Of\ Mind.flac | waveform -fg=#FF3300 -bg=#0099CC -alt=#FF9933 -fn fuzz -x 5 -y 2 > ~/waveform_fuzz.png

Apply a new set of colors, draw using a gradient pattern.

cat ~/Music/02\ -\ Peace\ Of\ Mind.flac | waveform -fg=#FF0000 -bg=#00FF00 -alt=#0000FF -fn gradient -x 5 -y 2 > ~/waveform_gradient.png

Apply a checkerboard color set, draw using a checkerboard pattern.

cat ~/Music/02\ -\ Peace\ Of\ Mind.flac | waveform -fg=#000000 -bg=#222222 -alt=#FFFFFF -fn checker -x 5 -y 2 > ~/waveform_checker.png

waveform's Issues

If possible, implement `io` interfaces

If possible, allow images to be generated, encoded, and directly streamed using io interfaces. This would enable images to be sent via HTTP as they are generated, etc.

Add waveform generation from []float64

Since []float64 is computed from reading samples anyway, allow direct generation by inputting []float64. Also add the option to retrieve the values used to compute a new waveform from a io.Reader, so that the same waveform can be quickly regenerated.

unlimited memory usage

when i generate with this;

img, err := waveform.Generate(out,waveform.Resolution(8000),bColor,fColor)

with a wave file with 8k samples per second, its fine, using 6MB memory in 0.01sec, but

img, err := waveform.Generate(out,waveform.Resolution(10000),bColor,fColor)

(ie when resolution slightly higher than the sample rate, )

i had to kill the process at 6GB mem usage.

Use "functional options" pattern instead of config structs

This will provide a much more simple interface, reduce the need to merge in default options, etc etc.

http://dave.cheney.net/2014/10/17/functional-options-for-friendly-apis

panic: EOF

Hi,

I'm using the latest commit of waveform: 6b28917
and Go 1.11 on 64-bit Arch Linux.

After building waveform and copying test.wav to the cmd/waveform directory:

$ ls -al
.rw-r----- 1,0k afr 14 sep.  11:36 README.md
.rw-r-----  47M afr 14 sep.  11:40 test.wav
.rwxr-x--- 3,0M afr 14 sep.  11:36 waveform*
.rw-r----- 4,8k afr 14 sep.  11:36 waveform.go
.rw-r-----    0 afr 14 sep.  11:40 waveform.png
$ file test.wav
test.wav: RIFF (little-endian) data, WAVE audio

When attempting to create a .png image, I get an error message:

$ cat test.wav | ./waveform -x 5 -y 2 > waveform.png
panic: EOF

goroutine 1 [running]:
main.main()
        /home/afr/clones/waveform/cmd/waveform/waveform.go:127 +0xc9b

I tried with several different .wav files.

Line 127 in waveform.go is here.

Compute method returns unknown format

Hi, I'm trying to get []float64 from Compute() method of Waveform, but it returns an error of "audio: unknown format", is there anything that I did wrong?

myAgi := agi.New()
myAgi.Init(nil)

fd, err := syscall.Open("/dev/fd/3", syscall.O_RDWR, 0755)  // open file descriptor
if err != nil {
	myAgi.Verbose(err)
	return
}

buf := make([]byte, 1024)

for {
	n, err := syscall.Read(fd, buf)  // read streaming audio
	if n > 0 {
		br := bytes.NewReader(buf[:n])

		wav, err := waveform.New(br)
		if err != nil {
			myAgi.Verbose(err)
		}

		value, err := wav.Compute()
		if err != nil {
			myAgi.Verbose(err)  // got an error here
		}

		// futher proccesing with value
	}

	if err != nil {
		myAgi.Verbose(err)
		return
	}
} 

something wrong with vertical scaling

this;

./waveform.a -resolution 1000 < wavTones.com.unregistred.sin_10Hz_-12dBFS_1s.wav > temp.png

(wav file from here; http://www.wavtones.com/functiongenerator.php github wont let me add wav files, but these both produce correct waveforms in audacity.)

produces this;

which is too big, -12db is 1/4 full scale.

and this, which is just full scale;

./waveform.a -resolution 1000 < wavTones.com.unregistred.sin_10Hz_0dBFS_1s.wav > temp.png

produces this? (completely white png)

Figure out a way to generate fixed-width images

Right now, image width is completely dependent on stream duration. In the future, it would be nice to be able to specify that the output image should be a certain width.