VoiceToScore

What is VoiceToScore?

The VoiceToScore system is a set of computer software which helps you to analyze recordings of voice reading poetry, to make a "poetical score," as shown in the above picture. Currently, the system is only for Windows PC.

A "poetical score" looks just like a music score, but they are different in two important ways:

  • Unlike in a music score, perpendicular bars in a poetical score (e. g., the single line between "now" and "and" in the above score, and the double line after "free" at the end of the verse) signify rests, or caesurae, in the reading.
  • A poetical score is descriptive; it describes, approximately, what is going on in the recording. A music score, on the other hand, is directive; it directs the musician how to play the piece.

The system can be downloaded from here.  (Owners of 32-bit Windows OS, download from here.) More detailed instruction for downloading is available here.)

To use the system for the first time, the first-time user needs to go through the following jobs:

  • Job 1. Preparation . Download the VoiceToScore system from our site to your computer according to directions given, and also download and install the LilyPond system.
  • Job 2. Practice. Use the voice data given in the "example" subfolder under the "data" folder and go through the process once.
  • Job 3. Production. Create a new data subfolder under the "data" folder, put the voice data you'd like to analyze there and go through the production process.

If you are familiar with the use of Praat system, even if you are a first-time user, you may skip Job 2, going directly from Job 1 to Job 3.

Job 2 and job 3 are identical after the preparation of the voice data and the poetry text data. The common parts of both Job2 and Job3 consist of 4 stages:

  • Stage 1. Input syllable borders and syllable labels into a .TextGrid file. See "Stage 1.pptx" and "Detailed Instructions for Stage 1.docx" under "documents" folder.
  • Stage 2. Input line numbers and stress/slack (S/W) to the TextGrid file. See "Stage 2.pptx" under "documents" folder.
  • Stage 3. From the TextGrid file, create a .ly file. See "Stage 3.docx" under "documents" folder.
  • Stage 4. Feed the .ly file to LilyPond program and obtain the poetical score. See "Stage 4.docx" under "documents" folder.
Analyzing a recording of a short poem can be done in a sitting, but for most poems a production job can take several hours or even days. The system is designed so that the job can be safely suspended anytime and re-started later.


The VoiceToScore system, being developed by Takeo Yamamoto, is available under Attribution-ShareAlike 4.0 International Licence. Some webpages on this site may be under construction and some of their content may refer to old versions. 

If anyone uses the system, reference should be given to the abstract of my paper at DH2015. The system, mostly in praat script, relies on Praat and LilyPond  systems. My thanks are due to the developers of those wonderful systems.The present website (for VoiceToScore Ver.0.8 and later) was written using BlueGriffon. Thanks are also due to Christpher Heng for CSS scripts.


©Takeo Yamamoto 2015-2018. Last updated 2018-3-3 for VoiceToScore Ver. 0.9. Please send inquiries to ty/atmark/tylab.org (please change "/atmark/" to "@".)