An open API service indexing awesome lists of open source software.

https://github.com/iann0036/deepcomposer-upload

Upload a single-track MIDI to the AWS DeepComposer service
https://github.com/iann0036/deepcomposer-upload

aws deepcomposer

Last synced: 6 months ago
JSON representation

Upload a single-track MIDI to the AWS DeepComposer service

Awesome Lists containing this project

README

          

# deepcomposer-upload

NPM version

> Upload a single-track MIDI to the AWS DeepComposer service

## Installation

```
npm i -g deepcomposer-upload
```

## Usage

Input MIDI files should have a tempo that remains static. Only the first 8 bars will be processed.

```
deepcomposer-upload -i input.mid -n nameoftrack -m genre-rock-1 -o output.mid
```

```
deepcomposer-upload \
-i input.mid \
-n nameoftrack \
-m ar-cnn-bach \
-o output.mid \
--max-percentage-removed 100 \
--max-notes-added 50 \
--sampling-iterations 100 \
--creative-risk 1.5
```

#### -i, --input-filename <filename> (Required)

The filename of the input MIDI file

#### -n, --sample-name <name>

The name of the sample, used to register in Music studio (required when uploading)

#### -m, --model-id <id>

The ID of the model used to generate new tracks, defaults to `genre-rock-1`

###### GAN-based

* `genre-rock-1`
* `genre-pop-1`
* `genre-jazz-1`
* `genre-joco-1`
* `genre-symphony-1`

###### Autoregressive-based

* `ar-cnn-bach`

#### -o, --output-filename <filename>

The name of the output filename of the generated MIDI file, if you wish to save it

#### --max-percentage-removed <number>

_Autoregressive only_

The maximum percentage of initial notes removed (0-100)

#### --max-notes-added <number>

_Autoregressive only_

The maximum notes to be added (50-1000)

#### --sampling-iterations <number>

_Autoregressive only_

The sampling iterations (0-100)

#### --creative-risk <number>

_Autoregressive only_

As risk increases, compositions will become more experimental (0.5-6)

#### --list-tracks

Lists tracks in the input MIDI file, and ignores uploading to the DeepComposer service

#### --track-number <number>

Selects the track number to process, as shown by `--list-tracks`