CITATION — REFERENCE ENTRY
Acoustic Chord Transcription and Key Extraction From Audio Using Key-Dependent HMMs Trained on Synthesized Audio — IEEE Transactions on Audio, Speech, and Language Processing
- Key
- lee2008-key-dependent-hmm
- Authors
- Lee, Kyogu; Slaney, Malcolm
- Issued
- 2008-2
- Type
- article-journal
- Container
- IEEE Transactions on Audio, Speech, and Language Processing
- Volume
- 16
- Issue
- 2
- Pages
- 291-301
Raw CSL JSON
{
"DOI": "10.1109/TASL.2007.914399",
"URL": "https://ccrma.stanford.edu/~kglee/pubs/klee-ieee-taslp08-print.pdf",
"page": "291-301",
"type": "article-journal",
"issue": "2",
"title": "Acoustic Chord Transcription and Key Extraction From Audio Using Key-Dependent HMMs Trained on Synthesized Audio",
"author": [
{
"given": "Kyogu",
"family": "Lee"
},
{
"given": "Malcolm",
"family": "Slaney"
}
],
"issued": {
"date-parts": [
[
2008,
2
]
]
},
"volume": "16",
"container-title": "IEEE Transactions on Audio, Speech, and Language Processing"
}
Claims
-
Lee and Slaney built one HMM for each of the 24 major and minor keys, estimating the key and the chord sequence together.
"We define major/minor keys for each pitch class, resulting in 24 different HMMs."
-
Lee and Slaney obtained training labels by automatic harmony analysis of symbolic music files and synthesized audio from the same files, giving acoustic features perfectly aligned with the labels.
"We avoid the extremely laborious task of human annotation of chord names and boundaries—which must be done to provide machine learning models with ground truth—by performing automatic harmony analysis on symbolic music files. In parallel, we synthesize audio from the same symbolic files and extract acoustic feature vectors which are in perfect alignment with the labels."
Available in