Мы очень рады показать вам наше исполнение известной русской песни "Катюша". Большое спасибо гречке (Грейс) для вокала и Вирджинии, которая играла на скрипке.

Катюша

Audio generated by training an LSTM neural network on John Coltrane's 1957 record, Blue Train. Created using Matt Vitelli's GRUV https://github.com/genekogan/GRUV

Blue Train LSTM

Audio produced by an LSTM neural network trained on Frederic Chopin's "Waltz in A Flat, Op. 69 No.1 (L'adieu)" played by Dinu Lipatti. 
Created with GRUV from Matt Vitelli: http://www.github.com/genekogan/GRUV

Frederic Chopin LSTM

Audio generated by training an LSTM neural network on a short sample of a vibraphone. Created mostly using Matt Vitelli's GRUV https://github.com/genekogan/GRUV

Original sample: http://github.com/processing/processing-sound/blob/master/examples/Soundfile/Sample/data/vibraphon.aiff

After the original generation with GRUV, I also applied a notch filter and reverb to smooth out some of the noise, and added a bass hum with SuperCollider:

({
	var freq, mul;
	freq=139.582;
	mul=[SinOsc.kr(80,2,0.1,0.2),SinOsc.kr(100,1,0.1,0.2),SinOsc.kr(90,3,0.1,0.2)];
	Mix.ar(SinOsc.ar([freq,freq/2,f/4],0,mul))!2
}.play)


Vibraphone LSTM

Some snippets of a guitar duet recorded 7/8/2011 with Matt Stauffer, Gene Kogan, and the soothing background of a fish tank filtration device.

Guitar duet improv + fish tank

wavenet trained on Giovanni Paisiello's Il Barbiere di Siviglia (The Barber of Seville).

made with tensorflow-wavenet https://github.com/ibab/tensorflow-wavenet.

learning rate 0.001, 25k steps, parameters: 
{
    "filter_width": 2,
    "sample_rate": 16000,
    "dilations": [1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
                  1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
                  1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
                  1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
                  1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
                  1, 2, 4, 8, 16, 32, 64],
    "residual_channels": 32,
    "dilation_channels": 32,
    "quantization_channels": 256,
    "skip_channels": 1024,
    "use_biases": true
}

Il Barbiere di Siviglia wavenet

Song "Amber" by 311 playing only the first beat of every bar, i.e. 123412341234 --> 111.  Thanks EchoNest API!

311 Amber, 1st beat of every bar

A collection of field recordings taken on a trip to Burma/Myanmar in Feb 2013.

0:00 bells in the last remaining Hindu temple of Rangoon/Yangon
0:20 market in the Muslim/Jewish quarter of Rangoon
0:40 a Buddhist prayer group along the length of a side street close to Sule Pagoda, Rangoon
1:25 a family huddles around an old Chinese movie on the bank of the Ayerwaddy river in Myitkyina, Kachin state
1:45 vendor selling food inside the train at a rest stop between Myitkyina and Hopin
2:15 a woman describing her shrine in her village along the northern bank of Lake Indawgyi
2:25 our boat guide at Lake Indawgyi playing with a baby
2:40 an oxcart waddles down the road, a Shan village around Lake Indawgyi
3:00 pig having breakfast around dawn in Lonton by Lake Indawgyi
4:05 listening to Frank Sinatra's On The Road to Mandalay from the back of a taxi in Mandalay
4:45 a resonating bell inside the old city walls of Bagan 
4:55 a group of Buddhist monks chanting from the top of Shwesandaw Pagoda in Bagan
6:15 crickets in the dead of night, Lonton, Lake Indawgyi, Kachin state

Burma compilation

Buddhist chant at sunrise from top of pagoda in Bagan. Collection of field recordings from Burma/Myanmar, Feb 2013.

Buddhist chant at sunrise from top of pagoda in Bagan

akhu chingangbam of imphal talkies jamming on ukulele

Imphal - Akhu jamming

a schoolbus carrying a catholic choir group singing 

Khonoma schoolbus choir

interviewing a sleeping pig near loktak lake, manipur

Sleeping pig

Tzilike, Nagaland after dark -- the crickets and cicadas in full swing

Tzilike at night

chants from a hindu brahmin living on majulie island, assam

Majulie Brahmin singing

the toy train between darjeeling and kurseong

Darjeeling toy train

Darjeeling Buddhist horns

Pre-Holi celebrations at New Jalpaiguri railway station 

Gene Kogan

Popular tracks