ScoreDraft is a music/singing synthesizer that provides a Python based score authoring interface.
Historically, there have been changes in the design goal of ScoreDraft. The software was originally intendend to be used for music experiments such as just intonation. For the initial design, I was intending to use text files as the means of authoring, i.e., the C++ program reads a text file, parse the notes and generates an audio file. However, after using Python as a script layer, it has shown very much benefit comparing to text parsing. So I moved quickly to Python.
I became interested in extending ScoreDraft for singing synthesis, simliar to Vocaloid and UTAU, but using script. Since then singing synthesis has become the focus of development. In the first stage, I implemented a simple voice synthesizing engine that generates a voice note by note from individual voice samples. This engine is called "KeLa", named after the famous video on Bilibili. In the second stage, I built a new engine makes use of voice-banks made for UTAU, including CVVC and VCV voice-banks. The new engine is called "UtauDraft". The name only indicates the intension to be compatible with UTAU voice banks.
I'm not sure where ScoreDraft will go in the furture. From here I'm building a big set of utilities to allow different kinds all waveforms to be generated and mixed together through scripts.
The following example shows how easily a piece of musical sound can be generated using ScoreDraft.
importScoreDraftfromScoreDraft.Notesimport*doc=ScoreDraft.Document()
seq=[do(),do(),so(),so(),la(),la(),so(5,96)]
doc.playNoteSeq(seq, ScoreDraft.Piano())
doc.mixDown('twinkle.wav')I'm intending to make all my source-code contribution to GitHub totally free to use.
ScoreDraft is now formally available under MIT license.
With
Dependecies
Source code under /ScoreDraft_local/SF2Instrument is based on TinySoundFont, Copyright (C) 2017 Bernhard Schelling (https://github.com/schellingb/TinySoundFont)
Exclusions
The above licenses only apply to the source code within the repository. Binaries under /python_test are not included in the license. Including (not limited to):
- Dlls from Qt5 are in LGPL license
- Audio samples from
https://freewavesamples.comhttp://utau.vocalover.com/
have their own licenses.
Users are on their own responsibility to properly use binaries under /python_test.
DON'T USE THE CONTENT OF THIS REPOSITORY TO DO ANY BAD THINGS. THE AUTHOR OF SCOREDRAFT DOES NOT HAVE ANY RESPONSIBILITY IN SUCH CASE. NO WARRANTY OF ANY KIND FROM THE AUTHOR.
First please check:
/python_test directory.
The above directory contains the whole Python layer of ScoreDraft. Since the core functions are implemented in C++, I also pre-deployed binaries built for Windows x64 and Linux Amd64 in that directory, so users of these systems can have a really quick evaluation without building ScoreDraft from source code.
These files are being dynamically updated as the development progresses. In that way I think there will be no need to make any formal releases. Audio sample files will be kept at small scale here. Users can extend that part themselves.
Sub-directories:
/python_test/ScoreDraft: the ScoreDraft package /python_test/ScoreDraft/Extensions: binary extensions
/python_test/ScoreDraft/InstrumentSamples: audio samples for InstrumentSampler extension
/python_test/ScoreDraft/PercussionSamples: audio samples for PercussionSampler extension
/python_test/ScoreDraft/KeLaSamples: audio samples for KeLa voice engine /python_test/ScoreDraft/UTAUVoice: deploy your UTAU voice-banks here, the UtauDraft voice engine will use
/python_test/zh_TTS: a simple Chinese TTS demo using pypinyin + ScoreDraft Examples:
/python_test/Hello.py: the simplest example
/python_test/piano_test.py: test of instrument sampler
/python_test/percussion_test.py: test of percussion sampler
/python_test/FlyMeToTheMoon.py: fly me to the moon in just-intonation
/python_test/FlyMeToTheMoon_eq.py: fly me to the moon in equal-temperment
/python_test/NaushikaRequiem.py: Naushika Requiem using a simple voice, with accompany
/python_test/KeLaTest.py: a test using the Jin-Ke-La voice bank
/python_test/GePi.py: a 10 minute rendering of the Pi song using GePing voice-bank
/python_test/GePi_unlimited.py: unlimited real-time generation/playback of the Pi song. /python_test/RapTest.py: test of the the Rap function by singing Chinese Peoms. Need the full-set of "GePing" voice-bank
/python_test/tang300.py: the database for RapTest.py
/python_test/CVVCTest.py: a test using a English CVVC voice-bank, specifically:
[kasaneteto, English Voicebank (CVVC) Voicebank](http://kasaneteto.jp/en/voicebank.html)
/python_test/CVVCTest2.py: a test using a CVVCChinese voice-bank, specifically:
[三色あやかCVVChinese V1.0](https://bowlroll.net/file/53297)
/python_test/VCVTest.py: a test using a Tsuro style Chinese VCV voice-bank, specifically:
[綰儿] (http://lindayana.lofter.com/waner)
/python_test/uta_test.py: a test using Japanese voice-banks, specifically:
a. the "uta" voice-bank that comes with UTAU
b. [三色あやか 連続音V2.0](https://bowlroll.net/file/69898)
/python_test/XiaYYTest.py: a test using XiaYuYao voice-bank, which is a special form of CVVCChinese. [XiaYuYao](http://www.voicemith.com/download.html)
Contents outside of python_test are basically C++ source-code files managed with CMake, which are used to build the binaries.
English introduction
http://scoredraft.org/index.php/introduction-to-scoredraft/
Chinese introduction (中文使用介绍）:
http://scoredraft.org/index.php/scoredraft-introduction-zh/
To build ScoreDraft from source-code, you need to install:
- CMake 3.0+
- Python3
- Qt5 (Needed by "QtPCMPlayer“ and "Meteor" extensions. You can remove them from /CMakeLists.txt if you don't need them. Then you don't need Qt5)
- CUDA (Optionally needed by UtauDraft, can be disabled by setting "USE_CUDA" to false in /UtauDraft/CMakeLists.txt)
Run CMake to generate makefiles/project files for your system and build. You are recommanded to:
- use the /build directory as you building directory
- use /python_test as your CMAKE_INSTALL_PREFIX