YIN, a fundamental frequency estimator for speech and music.

By A. de Cheveigné, Hideki Kawahara. 2002

2,301 citations recorded by the source corpus.

Subfields: Speech & Audio, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Speech & Audio

In Signal Processing

Citations in the corpus

Sample Efficient Adaptive Text-to-Speech (2018, methodology): “0 values were extracted using an YIN 0 tracker (De Cheveigné and Kawahara, 2002).”
Towards End-to-End Prosody Transfer for Expressive Speech Synthesis with Tacotron (2018, methodology): “All pitch and voicing metrics are computed using the output of the YIN (De Cheveigné & Kawahara, 2002) pitch tracking algorithm.”
WORLD: A Vocoder-Based High-Quality Speech Synthesis System for Real-Time Applications (2016, methodology): “Even though it is much faster than other algorithms such as YIN [21] and SWIPE [22], its estimation performance is not inferior to theirs [15].”
Flowtron: an Autoregressive Flow-based Generative Network for Text-to-Speech Synthesis (2020, methodology): “In Figure 4 we provide scatter plots of F 0 contours extracted with the YIN algorithm (De Cheveign´e & Kawahara, 2002), with minimum F 0 , maximum F 0 and harmonicity threshold equal to 80 Hz, 400 Hz and 0.3 respectively.”
Generating Diverse and Natural Text-to-Speech Samples Using a Quantized Fine-Grained VAE and Autoregressive Prosody Prior (2020, methodology): “The duration of a phoneme is represented by the number of frames where the decoder attention assigns maximum values to that phoneme, F0 is computed by the YIN pitch tracker [25], and the relative energy is the ratio of the average signal magnitude within a phoneme with the average magnitude of the e”
Neural Analysis and Synthesis: Reconstructing Speech from Self-Supervised Representations (2021, methodology): “Because the proposed feature is heavily inspired by the famous Yin algorithm [13], we refer to this feature as Yingram.”
Recognising realistic emotions and affect in speech: State of the art and lessons learnt from the first challenge (2011, methodology): “…using frame-based analysis: The speech signal is broken into overlapping frames and a pitch value is inferred from each segment mostly by autocorrelation (Rabiner, 1977) in its manifold variants and derivatives (Boersma, 1993; Cheveigne and Kawahara, 2002; Sood et al., 2004; Ding et al., 2006).”
Chunked Autoregressive GAN for Conditional Waveform Synthesis (2021, methodology): “We hypothesize that fully-connected layers are advantageous because they provide an inductive bias towards learning the autocorrelation function, which is a common operation used in DSP-based pitch estima-5 tion (De Cheveign´e & Kawahara, 2002; Rabiner & Schafer, 1978).”
StyleTTS: A Style-Based Generative Model for Natural and Diverse Text-to-Speech Synthesis (2022, neutral): “Our pitch extractor F is a JDC network [44], pre-trained on LibriSpeech with ground truth F0 estimated using YIN [45].”
Multiple fundamental frequency estimation based on harmonicity and spectral smoothness (2003, methodology): “The first method, YIN , is a state-of-the-art monophonic F0 estimator for speech and music signals [49].”
A pitch extraction algorithm tuned for automatic speech recognition (2014, methodology): “We selected three of the best-performing methods for further study; these were SAcC, Yin and getf0 (we did not consider YAAPT at this point because of its greater complexity; it is based on getf0).”
Towards Transfer Learning for End-to-End Speech Synthesis from Deep Pre-Trained Language Models (2019, methodology): “The pitch and voicing metrics are computed using the YIN [22] pitch tracking algorithm.”
Polynomial Chirplet Transform With Application to Instantaneous Frequency Estimation (2011, background): “The IF concept has played important roles for the study of a wide range of signals, from radar [5], sonar [6], biomedical engineering [7], seismic investigation [8], and speech and music[9] to automotive signals [10], where the IFs are often used to characterize important physical parameters of the ”
Mellotron: Multispeaker Expressive Voice Synthesis by Conditioning on Rhythm, Pitch and Global Style Tokens (2019, neutral): “Our pitch contours are automatically extracted using the Yin algorithm [10] with harmonicity thresholds between 0.1 and 0.25.”