andreasjansson / music-inpainting-bert

Music inpainting of melody and chords

Cold

Public
8.6K runs
T4
GitHub
License

Run with an API

Playground API Examples README Versions

Input

notes

*string

Shift + Return to add a new line

e2 f#4 | g.4 f#8 e4 | b2 b4 | b.2 | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | e.2e2 f#4 | g.4 f#8 e4 | b2 b4 | b.2 | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | e.2

Notes in tinynotation, with each bar separated by '|'. Use '?' for bars you want in-painted.

chords

*string

Shift + Return to add a new line

Em | Em | Em | Em | Am | Am | ? | ? | ? | ? | ? | ? | B7 | B7 | Em | EmEm | Em | Em | Em | Am | Am | ? | ? | ? | ? | ? | ? | B7 | B7 | Em | Em

Chords (one chord per bar), with each bar separated by '|'. Use '?' for bars you want in-painted.

time_signature

integer

Time signature

Default: 4

tempo

integer

(minimum: 60, maximum: 200)

Tempo (beats per minute)

Default: 120

sample_width

integer

(minimum: 1, maximum: 200)

Number of potential predictions to sample from. The higher, the more chaotic the output.

Default: 10

seed

integer

Random seed, -1 for random

Default: -1

Run this model in Node.js with one line of code:

npx create-replicate --model=andreasjansson/music-inpainting-bert

or set up a project from scratch

Install Replicate’s Node.js client library:

npm install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import and set up the client:

import Replicate from "replicate";

const replicate = new Replicate({
  auth: process.env.REPLICATE_API_TOKEN,
});

Run andreasjansson/music-inpainting-bert using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "andreasjansson/music-inpainting-bert:58bdc2073c9c07abcc4200fe808e15b1a555dbb1390e70f5daa6b3d81bd11fb1",
  {
    input: {
      seed: -1,
      notes: "e2 f#4 | g.4 f#8 e4 | b2 b4 | b.2 | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | e.2",
      tempo: 100,
      chords: "Em | Em | Em | Em | Am | Am | ? | ? | ? | ? | ? | ? | B7 | B7 | Em | Em",
      sample_width: 80,
      time_signature: 3
    }
  }
);

console.log(output);

To learn more, take a look at the guide on getting started with Node.js.

Install Replicate’s Python client library:

pip install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import the client:

import replicate

Run andreasjansson/music-inpainting-bert using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "andreasjansson/music-inpainting-bert:58bdc2073c9c07abcc4200fe808e15b1a555dbb1390e70f5daa6b3d81bd11fb1",
    input={
        "seed": -1,
        "notes": "e2 f#4 | g.4 f#8 e4 | b2 b4 | b.2 | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | e.2",
        "tempo": 100,
        "chords": "Em | Em | Em | Em | Am | Am | ? | ? | ? | ? | ? | ? | B7 | B7 | Em | Em",
        "sample_width": 80,
        "time_signature": 3
    }
)
print(output)

To learn more, take a look at the guide on getting started with Python.

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Run andreasjansson/music-inpainting-bert using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

curl -s -X POST \
  -H "Authorization: Bearer $REPLICATE_API_TOKEN" \
  -H "Content-Type: application/json" \
  -H "Prefer: wait" \
  -d $'{
    "version": "andreasjansson/music-inpainting-bert:58bdc2073c9c07abcc4200fe808e15b1a555dbb1390e70f5daa6b3d81bd11fb1",
    "input": {
      "seed": -1,
      "notes": "e2 f#4 | g.4 f#8 e4 | b2 b4 | b.2 | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | e.2",
      "tempo": 100,
      "chords": "Em | Em | Em | Em | Am | Am | ? | ? | ? | ? | ? | ? | B7 | B7 | Em | Em",
      "sample_width": 80,
      "time_signature": 3
    }
  }' \
  https://api.replicate.com/v1/predictions

To learn more, take a look at Replicate’s HTTP API reference docs.

Output

mp3

Video Player is loading.

Current Time 00:00:000

Duration 00:00:000

Loaded: 0%

Stream Type LIVE

Remaining Time 00:00:000

midi

out.midi

score

{
  "completed_at": "2022-05-03T19:42:15.728031Z",
  "created_at": "2022-05-03T19:42:08.098990Z",
  "data_removed": false,
  "error": null,
  "id": "6d4inmhz6vadrpbq6g7zgmrbu4",
  "input": {
    "seed": -1,
    "notes": "e2 f#4 | g.4 f#8 e4 | b2 b4 | b.2 | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | ? | e.2",
    "tempo": "100",
    "chords": "Em | Em | Em | Em | Am | Am | ? | ? | ? | ? | ? | ? | B7 | B7 | Em | Em",
    "sample_width": "80",
    "time_signature": "3"
  },
  "logs": "Using seed: 23529\n0.0 80\n0.0 10\n0.1 12\n0.1 8\n0.2 9\n0.2 6\n0.3 7\n0.3 6\n0.4 6\n0.4 5\n0.5 5\n0.5 4\n0.6 4\n0.6 4\n0.7 3\n0.7 3\n0.8 3\n0.8 2\n0.9 2\n0.9 2\n0:0 (0/1 E4) Em/E 0\n0:1  Em/E 0\n1:2 (0/1 F#4) Em/E 0\n1:0 (0/1 G4) Em/E 0\n1:1 (1/2 F#4) Em/E 0\n2:2 (0/1 E4) Em/E 0\n2:0 (0/1 B4) Em/E 0\n2:1  Em/E 0\n3:2 (0/1 B4) Em/E 0\n3:0 (0/1 B4) Em/E 0\n3:1  Em/E 0\n4:2  Em/E 0\n4:0 (0/1 G4) Am/A 7\n4:1 (0/1 F#4) Am/A 4\n5:2 (0/1 E4) Am/A 6\n5:0 (0/1 G4) Am/A 3\n5:1 (0/1 F#4) Am/A 5\n6:2 (0/1 E4) Am/A 1\n6:0 (0/1 F#4) Em/E 7\n6:1 (0/1 E4) Em/E 2\n7:2 (0/1 G4) Em/E 5\n7:0 (0/1 B4) Em/E 4\n7:1  Em/E 8\n8:2  Em/E 8\n8:0 (0/1 B4) Em/E 3\n8:1  Em/E 5\n9:2 (0/1 B4), (1/2 B4) Em/E 6\n9:0 (0/1 B4) Am/A 2\n9:1 (0/1 A4) Am/A 6\n10:2 (0/1 G4) Am/A 7\n10:0 (0/1 B4), (1/2 G4) Em/E 1\n10:1 (0/1 E4) Em/E 4\n11:2  Em/E 2\n11:0 (0/1 E4) Em/E 2\n11:1 (0/1 E-4) Em/E 9\n12:2 (0/1 D4) Em/E 1\n12:0 (0/1 E-4) B7 3\n12:1  B7 4\n13:2 (0/1 E-4), (1/2 E-4) B7 7\n13:0 (0/1 F#4) B7 5\n13:1 (0/1 E-4) B7 3\n14:2 (1/2 B3) B7 1\n14:0 (0/1 E4) Em/E 6\n14:1  Em/E 8\n15:2  Em/E 8\n15:0 (0/1 E4) Em/E 0\n46: Invalid pattern: 0\n0 pattern\nffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers\n  built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)\n  configuration: --prefix=/usr --extra-version=0ubuntu0.2 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --enable-gpl --disable-stripping --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librubberband --enable-librsvg --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libopencv --enable-libx264 --enable-shared\n  libavutil      55. 78.100 / 55. 78.100\n  libavcodec     57.107.100 / 57.107.100\n  libavformat    57. 83.100 / 57. 83.100\n  libavdevice    57. 10.100 / 57. 10.100\n  libavfilter     6.107.100 /  6.107.100\n  libavresample   3.  7.  0 /  3.  7.  0\n  libswscale      4.  8.100 /  4.  8.100\n  libswresample   2.  9.100 /  2.  9.100\n  libpostproc    54.  7.100 / 54.  7.100\nGuessed Channel Layout for Input Stream #0.0 : mono\nInput #0, wav, from '/tmp/tmp_fuotcp4/out.wav':\n  Duration: 00:00:29.20, bitrate: 2822 kb/s\n    Stream #0:0: Audio: pcm_f64le ([3][0][0][0] / 0x0003), 44100 Hz, mono, dbl, 2822 kb/s\nStream mapping:\n  Stream #0:0 -> #0:0 (pcm_f64le (native) -> mp3 (libmp3lame))\nPress [q] to stop, [?] for help\nOutput #0, mp3, to '/tmp/tmp_fuotcp4/out.mp3':\n  Metadata:\n    TSSE            : Lavf57.83.100\n    Stream #0:0: Audio: mp3 (libmp3lame), 44100 Hz, mono, fltp\n    Metadata:\n      encoder         : Lavc57.107.100 libmp3lame\nsize=     226kB time=00:00:28.83 bitrate=  64.1kbits/s speed=68.2x\nvideo:0kB audio:226kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.098302%\nLY output to `/tmp/tmp_fuotcp4/score.ly'...\nChanging working directory to: `/tmp/tmp_fuotcp4'\nProcessing `/tmp/tmp_fuotcp4/score.ly'\nParsing...\nInterpreting music...[8][16][16]\nPreprocessing graphical objects...\nInterpreting music...\nMIDI output to `score.midi'...\nFinding the ideal number of pages...\nFitting music on 1 page...\nDrawing systems...\nLayout output to `score.ps'...\nConverting to PNG...\nSuccess: compilation successfully completed",
  "metrics": {
    "predict_time": 7.443402,
    "total_time": 7.629041
  },
  "output": {
    "mp3": "https://replicate.delivery/mgxm/967ac956-4d23-4b2e-8d25-70165285ced9/out.mp3",
    "midi": "https://replicate.delivery/mgxm/380e6ecf-b993-4e2e-9743-6efd965fca28/out.midi",
    "score": "https://replicate.delivery/mgxm/fd0dd91d-939c-447d-b252-83d920918c22/score.png"
  },
  "started_at": "2022-05-03T19:42:08.284629Z",
  "status": "succeeded",
  "urls": {
    "get": "https://api.replicate.com/v1/predictions/6d4inmhz6vadrpbq6g7zgmrbu4",
    "cancel": "https://api.replicate.com/v1/predictions/6d4inmhz6vadrpbq6g7zgmrbu4/cancel"
  },
  "version": "58bdc2073c9c07abcc4200fe808e15b1a555dbb1390e70f5daa6b3d81bd11fb1"
}

Generated in

7.5 seconds

Tweak itReport

Using seed: 23529
0.0 80
0.0 10
0.1 12
0.1 8
0.2 9
0.2 6
0.3 7
0.3 6
0.4 6
0.4 5
0.5 5
0.5 4
0.6 4
0.6 4
0.7 3
0.7 3
0.8 3
0.8 2
0.9 2
0.9 2
0:0 (0/1 E4) Em/E 0
0:1  Em/E 0
1:2 (0/1 F#4) Em/E 0
1:0 (0/1 G4) Em/E 0
1:1 (1/2 F#4) Em/E 0
2:2 (0/1 E4) Em/E 0
2:0 (0/1 B4) Em/E 0
2:1  Em/E 0
3:2 (0/1 B4) Em/E 0
3:0 (0/1 B4) Em/E 0
3:1  Em/E 0
4:2  Em/E 0
4:0 (0/1 G4) Am/A 7
4:1 (0/1 F#4) Am/A 4
5:2 (0/1 E4) Am/A 6
5:0 (0/1 G4) Am/A 3
5:1 (0/1 F#4) Am/A 5
6:2 (0/1 E4) Am/A 1
6:0 (0/1 F#4) Em/E 7
6:1 (0/1 E4) Em/E 2
7:2 (0/1 G4) Em/E 5
7:0 (0/1 B4) Em/E 4
7:1  Em/E 8
8:2  Em/E 8
8:0 (0/1 B4) Em/E 3
8:1  Em/E 5
9:2 (0/1 B4), (1/2 B4) Em/E 6
9:0 (0/1 B4) Am/A 2
9:1 (0/1 A4) Am/A 6
10:2 (0/1 G4) Am/A 7
10:0 (0/1 B4), (1/2 G4) Em/E 1
10:1 (0/1 E4) Em/E 4
11:2  Em/E 2
11:0 (0/1 E4) Em/E 2
11:1 (0/1 E-4) Em/E 9
12:2 (0/1 D4) Em/E 1
12:0 (0/1 E-4) B7 3
12:1  B7 4
13:2 (0/1 E-4), (1/2 E-4) B7 7
13:0 (0/1 F#4) B7 5
13:1 (0/1 E-4) B7 3
14:2 (1/2 B3) B7 1
14:0 (0/1 E4) Em/E 6
14:1  Em/E 8
15:2  Em/E 8
15:0 (0/1 E4) Em/E 0
46: Invalid pattern: 0
0 pattern
ffmpeg version 3.4.8-0ubuntu0.2 Copyright (c) 2000-2020 the FFmpeg developers
  built with gcc 7 (Ubuntu 7.5.0-3ubuntu1~18.04)
  configuration: --prefix=/usr --extra-version=0ubuntu0.2 --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --enable-gpl --disable-stripping --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librubberband --enable-librsvg --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libopencv --enable-libx264 --enable-shared
  libavutil      55. 78.100 / 55. 78.100
  libavcodec     57.107.100 / 57.107.100
  libavformat    57. 83.100 / 57. 83.100
  libavdevice    57. 10.100 / 57. 10.100
  libavfilter     6.107.100 /  6.107.100
  libavresample   3.  7.  0 /  3.  7.  0
  libswscale      4.  8.100 /  4.  8.100
  libswresample   2.  9.100 /  2.  9.100
  libpostproc    54.  7.100 / 54.  7.100
Guessed Channel Layout for Input Stream #0.0 : mono
Input #0, wav, from '/tmp/tmp_fuotcp4/out.wav':
  Duration: 00:00:29.20, bitrate: 2822 kb/s
    Stream #0:0: Audio: pcm_f64le ([3][0][0][0] / 0x0003), 44100 Hz, mono, dbl, 2822 kb/s
Stream mapping:
  Stream #0:0 -> #0:0 (pcm_f64le (native) -> mp3 (libmp3lame))
Press [q] to stop, [?] for help
Output #0, mp3, to '/tmp/tmp_fuotcp4/out.mp3':
  Metadata:
    TSSE            : Lavf57.83.100
    Stream #0:0: Audio: mp3 (libmp3lame), 44100 Hz, mono, fltp
    Metadata:
      encoder         : Lavc57.107.100 libmp3lame
size=     226kB time=00:00:28.83 bitrate=  64.1kbits/s speed=68.2x
video:0kB audio:226kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.098302%
LY output to `/tmp/tmp_fuotcp4/score.ly'...
Changing working directory to: `/tmp/tmp_fuotcp4'
Processing `/tmp/tmp_fuotcp4/score.ly'
Parsing...
Interpreting music...[8][16][16]
Preprocessing graphical objects...
Interpreting music...
MIDI output to `score.midi'...
Finding the ideal number of pages...
Fitting music on 1 page...
Drawing systems...
Layout output to `score.ps'...
Converting to PNG...
Success: compilation successfully completed

Examples

View more examples

Run time and cost

This model costs approximately $0.0014 to run on Replicate, or 714 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia T4 GPU hardware. Predictions typically complete within 7 seconds. The predict time for this model varies significantly based on the inputs.

Readme

Music inpainting with BERT

This model generates monophonic melody and chords jointly, by filling in missing pieces.

In the notes and chords inputs, you can use ? to let the machine generate notes and chords for a bar at a time. See and tweak the examples to get a better idea of how it works.

Notes are input in TinyNotation format.

This model was trained on the Wikifonia dataset.