The binaries and example recordings picked for Dunya-makam demo
This repository includes audio-score alignment binaries compiled in MATLAB 2015a, which are be used in Dunya Ottoman-Turkish makam. Using the related SymbTr score, you can do the following:
- Joint tonic, tempo, tuning extraction
- Section, phrase linking
- Note-level alignment
Examples:
- Tonic, Tempo, Tuning
./run_mcr.sh extractTonicTempoTuning score.txt scoreMetadata.json audio.wav audioMelody.mat outputFolder
- Sections, Notes
./run_mcr.sh alignAudioScore score.txt scoreMetadata.json structureModel.mat audio.wav audioMelody.mat audioTonic.json audioTempo.json audioTuning.json outputFolder
We also supply 10 challenging audio-score examples with complete analysis & alignment. These arew used to demonstrate the strengths/weaknesses of our audio-score alignment methodology.
Notes:
- extractTonicTempoTuning requires scoreMetadata.json and audioMelody.mat computed by metadata.py and pitch.py, respectively.
- In addition to scoreMetadata.json and audioMelody.mat, alignAudioScore requires also the tonic.json, tempo.json and tuning.json files generated by extractTonicTempoTuning.
- The tempo, tuning and sectionModel inputs of the alignAudioScore are currently not used internally and they can be omitted by supplying an empty string ('').
- alignAudioScore initially does partial-alignment of the sections given in scoreMetadata.json. Make sure that the section information is entered correctly.
- Always supply wav files to alignAudioScore in order to avoid time mismatches in the note-level alignment due to mp3 decoders.