Methods › General › Attention Mechanisms › Location Sensitive Attention
Location Sensitive Attention
Introduced by Jan Chorowski et al. in Attention-Based Models for Speech Recognition
archive 2025-07-28 Description, source and code snippet are the archive's method entry.
Location Sensitive Attention is an attention mechanism that extends the additive attention mechanism to use cumulative attention weights from previous decoder time steps as an additional feature. This encourages the model to move forward consistently through the input, mitigating potential failure modes where some subsequences are repeated or ignored by the decoder.
Starting with additive attention where h is a sequential representation from a BiRNN encoder and sᵢ₋₁ is the (i − 1)-th state of a recurrent neural network (e.g. a LSTM or GRU):
e_(i, j) = wᵀtanh(Wsᵢ₋₁ + Vhⱼ + b)
where w and b are vectors, W and V are matrices. We extend this to be location-aware by making it take into account the alignment produced at the previous step. First, we extract k vectors f_(i,j) ∈ℝᵏ for every position j of the previous alignment αᵢ₋₁ by convolving it with a matrix F ∈R^(k×r):
fᵢ = F ∗ αᵢ₋₁
These additional vectors f_(i,j) are then used by the scoring mechanism e_(i,j):
e_(i,j) = wᵀtanh(Wsᵢ₋₁ + Vhⱼ + Uf_(i,j) + b)
Papers archive 2025-07-28
25 shown of 25, newest first. Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code; it is per sample and not a correctness claim.
-
Training Universal Vocoders with Feature Smoothing-Based Augmentation Methods for High-Quality TTS Systems 4 Sep 2024 · 0 repositories · arXiv:2409.02517
-
An overview of text-to-speech systems and media applications 22 Oct 2023 · 0 repositories · arXiv:2310.14301
-
Energy-Based Models For Speech Synthesis 19 Oct 2023 · 0 repositories · arXiv:2310.12765
-
Multilingual Text-to-Speech Synthesis for Turkic Languages Using Transliteration 25 May 2023 · 1 repository · arXiv:2305.15749
-
ArmanTTS single-speaker Persian dataset 7 Apr 2023 · 0 repositories · arXiv:2304.03585
-
Facial Landmark Predictions with Applications to Metaverse 29 Sep 2022 · 1 repository · arXiv:2209.14698
-
Zero-Shot Long-Form Voice Cloning with Dynamic Convolution Attention 25 Jan 2022 · 0 repositories · arXiv:2201.10375
-
ITAcotron 2: Transfering English Speech Synthesis Architectures and Speech Features to Italian 1 Nov 2021 · 1 repository
-
Neural Sequence-to-Sequence Speech Synthesis Using a Hidden Semi-Markov Model Based Structured Attention Mechanism 31 Aug 2021 · 0 repositories · arXiv:2108.13985
-
Neural HMMs are all you need (for high-quality attention-free TTS) 30 Aug 2021 · 2 repositories · arXiv:2108.13320
-
Ctrl-P: Temporal Control of Prosodic Variation for Speech Synthesis 15 Jun 2021 · 0 repositories · arXiv:2106.08352
-
VARA-TTS: Non-Autoregressive Text-to-Speech Synthesis based on Very Deep VAE with Residual Attention 12 Feb 2021 · 0 repositories · arXiv:2102.06431
-
Bidirectional Variational Inference for Non-Autoregressive Text-to-Speech 1 Jan 2021 · 1 repository
-
Using previous acoustic context to improve Text-to-Speech synthesis 7 Dec 2020 · 0 repositories · arXiv:2012.03763
-
Learning Speaker Embedding from Text-to-Speech 21 Oct 2020 · 1 repository · arXiv:2010.11221
-
Non-Attentive Tacotron: Robust and Controllable Neural TTS Synthesis Including Unsupervised Duration Modeling 8 Oct 2020 · 6 repositories · arXiv:2010.04301Syntology ran 0 of 7 samples · 7 unverified
-
SpeedySpeech: Efficient Neural Speech Synthesis 9 Aug 2020 · 3 repositories · arXiv:2008.03802Syntology ran 1 of 1 samples · 0 unverified
-
One Model, Many Languages: Meta-learning for Multilingual Text-to-Speech 3 Aug 2020 · 1 repository · arXiv:2008.00768
-
Flowtron: an Autoregressive Flow-based Generative Network for Text-to-Speech Synthesis 12 May 2020 · 3 repositories · arXiv:2005.05957Syntology ran 4 of 19 samples · 15 unverified
-
Fully-hierarchical fine-grained prosody modeling for interpretable speech synthesis 6 Feb 2020 · 0 repositories · arXiv:2002.03785
-
WaveTTS: Tacotron-based TTS with Joint Time-Frequency Domain Loss 2 Feb 2020 · 0 repositories · arXiv:2002.00417
-
Probing the phonetic and phonological knowledge of tones in Mandarin TTS models 23 Dec 2019 · 1 repository · arXiv:1912.10915
-
Mellotron: Multispeaker expressive voice synthesis by conditioning on rhythm, pitch and global style tokens 26 Oct 2019 · 5 repositories · arXiv:1910.11997
-
Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictions 16 Dec 2017 · 33 repositories · arXiv:1712.05884Syntology ran 7 of 7 samples · 0 unverified · 2 pointer-only (licence)
-
Attention-Based Models for Speech Recognition 24 Jun 2015 · 14 repositories · arXiv:1506.07503
Tasks archive 2025-07-28
20 shown of 29 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.
Usage over time archive 2025-07-28
Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).
Categories archive 2025-07-28
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections