SpecTNT: a Time-Frequency Transformer for Music Audio.
Wei Tsung LuJu-Chiang WangMinz WonKeunwoo ChoiXuchen SongPublished in: ISMIR (2021)
Keyphrases
- audio signals
- music score
- music information retrieval
- audio features
- music scores
- music genre classification
- audio signal
- audio recordings
- music retrieval
- signal processing
- audio content
- automatic music genre classification
- speech music discrimination
- music collections
- digital audio
- digital music
- audio files
- fuzzy logic
- genre classification
- content based music retrieval
- musical instruments
- polyphonic music
- acoustic features
- fault diagnosis
- visual features
- multimedia
- computer music
- low level
- visual information
- audio video
- frequency domain
- multiresolution
- gaussian mixture model
- multimedia databases
- hidden markov models
- signal analysis
- music recommendation
- subband
- soccer video
- distribution network
- audio visual
- visual data
- expert systems
- short time fourier transform
- video sequences
- power transformers