Instrument recognition in polyphonic music
read more
Citations
Instrument identification in polyphonic music: feature weighting to minimize influence of sound overlaps
Dynamic Spectral Envelope Modeling for Timbre Analysis of Musical Instrument Sounds
Polyphonic Instrument Recognition Using Spectral Clustering.
Pitch Detection in Polyphonic Music using Instrument Tone Models
From Sparse Models to Timbre Learning: New Methods for Musical Source Separation
References
Robust text-independent speaker identification using Gaussian mixture speaker models
RWC Music Database: Popular, Classical, and Jazz Music Databases
A real-time music-scene-description system: predominant-F0 estimation for detecting melody and bass lines in real-world audio signals
Comparative study of techniques for large-scale feature selection
Sound-source recognition: a theory and computational model
Related Papers (5)
Musical instrument classification and duet analysis employing music information retrieval techniques
Frequently Asked Questions (8)
Q2. What are the future works in "Instrument recognition in polyphonic music" ?
Future work will be dedicated to the assessment of the proposed methodology on more varied musical genre and instrumentation.
Q3. How can the authors use the proposed system in a music segmentation task?
In fact, since short-time decisions can be taken (≈ 2s), the proposed system can be easily employed in a task of segmentation of music duos, trios and quartets.
Q4. What is the use of the number of instruments in the audio signal?
Another application of this work is the detection of the number of instruments or sources involved in the audio signal, which can be very helpful for source separation tasks.
Q5. What is the probability of a positive vote for a class?
For each pair of classes {Ωi, Ωj}, a positive vote is counted for the class Ωi at time t ifp(xt|Ωi) > p(xt|Ωj) (2) where xt is the test feature vector observed at time t and (p(xt|Ωk))k=i,j is the class-conditional probability of xt, which is modeled as a GMM.
Q6. What was the common reason for the recognition of musical instruments?
In fact, recognition was often related to a source separation task requiring the knowledge or estimation of the pitches of the different notes [5, 6, 7].
Q7. Why is a new feature set derived?
Since pitch estimation has to be avoided, a new feature set is derived to roughly evaluate the power distribution among the different harmonics [4].
Q8. How did Eggink & Brown propose a system for identifying 2 instruments simultaneously?
Using realistic musical recordings, Eggink & Brown proposed a system based on a missing feature approach [8] capable of identifying 2 instruments playing simultaneously.