Query by humming with the VocalSearch system

European Signal Processing …, 2009

We propose a new algorithm for onset detection in hummed queries for QBH applications. The algorithm uses a modified version of a popular loudness model for human hearing to identify onsets in hums. We also propose the use of a local minimum function to identify onsets better. A subband based sone scale processing that is advantageous for a simple implementation is used. On an annotated database of syllabic and natural hums, the algorithm identifies onsets correctly on an average 90% of the time with only 6% false positives. The features used in this algorithm can be used in conjunction with other feature / decision fusion based onset detection systems.

, 2 neha_chhatre@yahoo.co.in, 3 namratasurendranath@yahoo.com, 4 mann_k_0905@yahoo.co.in Abstract— A Query by Humming system allows the user to find a song by humming part of the tune. No musical training is needed. The idea is simple: you hum into the microphone, the computer records the hum and extracts certain features corresponding to the melody and rhythm characteristics, and it then compares the features to the features of the songs in the database. Finally it returns a ranked list of the songs or song segments most similar to the humming. The goal is to build a reliable and efficient large-scale system that collects thousands of songs and responds in seconds.

Rangsit University, 2011

To improve Query-by-Humming, in this paper, we propose a note segmentation by humming sound method, melody contour extraction technique, and new normalization methods. The noise interference from both the environment and acquisition instrument is the critical issues in humming sound. The query problems about variation of pitch and timing are because most users are not professional singers. The advantage of the note segmentation by humming method is it can separate the sound and silent parts from each other through the process. The melody contour extraction can reduce noise resulting in pitch smoothing. Our approach starts from pre-processing by using features for note segmentation by a humming sound. The process consists of three steps as follows: Firstly, the pitch is extracted from the humming sound by Subharmonic-to-Harmonic Ratio (SHR). Afterwards, we used various new normalization methods, including melody contour extraction, for scaling and noise robust. Finally, Dynamic Time Warping (DTW) is applied to the melody contour, for similarity of measurement between the humming sound and the melody sequence. Comparing our proposed technique and the traditional method, the results show that our proposed techniques can perform more effectively.

Proceedings of the 2nd ACM/IEEE-CS joint …, 2002

Query-by-humming systems search a database of music for good matches to a sung, hummed, or whistled melody. Errors in transcription and variations in pitch and tempo can cause substantial mismatch between queries and targets. Thus, algorithms for measuring melodic similarity in query-byhumming systems should be robust. We compare several variations of search algorithms in an effort to improve search precision. In particular, we describe a new frame-based algorithm that significantly outperforms note-by-note algorithms in tests using sung queries and a database of MIDI-encoded music.

Technology, 2007

Query-by-Humming systems offer content-based searching for melodies and require no special musical training or knowledge. Many such systems have been built, but there has not been much useful evaluation and comparison in the literature due to the lack of shared databases and queries. The MUSART project testbed allows various search algorithms to be compared using a shared framework that automatically runs experiments and summarizes results. Using this testbed, we compared algorithms based on string alignment, melodic contour matching, a hidden Markov model, n-grams, and CubyHum. Retrieval performance is very sensitive to distance functions and the representation of pitch and rhythm, which raises questions about some previously published conclusions. Some algorithms are particularly sensitive to the quality of queries. Our queries, which are taken from human subjects in a fairly realistic setting, are quite difficult, especially for n-gram models. Finally, simulations on query-byhumming performance as a function of database size indicate that retrieval performance falls only slowly as the database size increases.

Pattern Recognition Letters, 2014

Singing or humming to a music search engine is an appealing multimodal interaction paradigm, particularly for small sized portable devices that are ubiquitous nowadays. The aim of this work is to overcome the main shortcoming of the existing query-by-humming (QBH) systems: their lack of scalability in terms of the difficulty of automatically extending the database of melodies from audio recordings. A method is proposed to extract the singing voice melody from polyphonic music providing the necessary information to index it as an element in the database. The search of a query pattern in the database is carried out combining note sequence matching and pitch time series alignment. A prototype system was developed and experiments are carried out pursuing a fair comparison between manual and automatic expansion of the database. In the light of the obtained performance (85% in the top-10), which is encouraging given the results reported to date, this can be considered a proof of concept that validates the approach.

Many state-of-the-art query-by-humming (QBH) systems restrict the hummed query to be in isolated syllables for easy note segmentation. However, it is observed that users often prefer natural humming. This work addresses note onset detection for natural hum, usually considered difficult to segment. The acoustic characteristics of naturally hummed signals are studied and features useful to note onset detection are proposed. Pitch and energy features are combined to obtain superior note segmentation. Performance results on note onset detection as well as retrieval in an actual QBH system are reported.

The International journal of Multimedia & Its Applications, 2012

Query by Singing/Humming (QBSH) is a Music Information Retrieval (MIR) system with small audio excerpt as query. The rising availability of digital music stipulates effective music retrieval methods. Further, MIR systems support content based searching for music and requires no musical acquaintance. Current work on QBSH focuses mainly on melody features such as pitch, rhythm, note etc., size of databases, response time, score matching and search algorithms. Even though a variety of QBSH techniques are proposed, there is a dearth of work to analyze QBSH through query excerption. Here, we present an analysis that works on QBSH through query excerpt. To substantiate a series of experiments are conducted with the help of Mel-Frequency Cepstral Coefficients (MFCC), Linear Predictive Coefficients (LPC) and Linear Predictive Cepstral Coefficients (LPCC) to portray the robustness of the knowledge representation. Proposed experiments attempt to reveal that retrieval performance as well as precision diminishes in the snail phase with the growing database size.

Proceedings of the 1st ACM workshop on Audio and music computing multimedia - AMCMM '06, 2006

Transcription from audio to musical representation is a challenging problem for Query by Humming (QBH) systems. In this paper, we propose a two step note transcription process consisting of an algorithm that uses a speech recognizer for note segmentation followed by signal processing for robust location and capture of pitch and duration in the humming audio input. In contrast to most Hidden Markov Model based approaches to QBH systems that model and classify humming into a single universal model, we designed a flexible speech recognizer that allows different types of humming sounds in the input for providing efficient and accurate note segmentation and transcription. We use novel statistical energy and pitch analyses to correct potential insertion and deletion errors to augment the system's performance, and evaluate our algorithm with precision and recall tests. Using a large database previously amassed, we test various system configurations, providing results for note segmentation with and without the proposed augmentations. The augmented system robustly recognizes the location of humming notes with a precision and recall F measure of 0.84. As a second validation, we use the results of the transcription system in melody retrieval and found, for a database of 1000 melodies, a 76% retrieval accuracy with automatically extracted queries, and a 83% retrieval performance with manually transcribed queries. Sensitivity analysis shows that, while it is possible to locate the position of the hummed notes accurately, incorrect segmentation results can have a negative effect in the retrieval performance of the QBH system.