Mondegreen: A Post-Processing Solution to Speech Recognition Error Correction for Voice Search Queries

Sodhi, Sukhdeep; Chio, Ellie Ka-In; Jash, Ambarish; Ontañón, Santiago; Apte, Ajit; Kumar, Ankit; Jeje, Ayooluwakunmi; Kuzmin, Dima; Fung, Harry; Cheng, Heng-Tze; Effrat, Jon; Bali, Tarush; Jindal, Nihar; Cao, Pei; Singh, Sarvjeet; Zhou, Shuli; Khan, Tameen; Wankhede, Amol; Alzantot, Moustafa; Wu, Allen C.-H.; Chandra, Tushar

doi:10.1145/3447548.3467156

Cited by 3 publications

(2 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several prior studies have investigated the QR problem in a non-personalized context. Statistical QR models have been deployed in Alexa and Google voice search (Sodhi et al, 2021). In their seminal work, apply an Absorbing Markov Chain (AMC) model as a collaborative filtering mechanism to mine reformulation patterns from sequences of user queries.…”

Section: Related Workmentioning

confidence: 99%

PAIGE: Personalized Adaptive Interactions Graph Encoder for Query Rewriting in Dialogue Systems

Biś,

Gupta,

Hao

et al. 2022

Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing: Industry Track

View full text Add to dashboard Cite

Unexpected responses or repeated clarification questions from conversational agents detract from the users' experience with technology meant to streamline their daily tasks. To reduce these frictions, Query Rewriting (QR) techniques replace transcripts of faulty queries with alternatives that lead to responses that satisfy the users' needs. Despite their successes, existing QR approaches are limited in their ability to fix queries that require considering users' personal preferences. We improve QR by proposing Personalized Adaptive Interactions Graph Encoder (PAIGE). PAIGE is the first QR architecture that jointly models user's affinities and query semantics end-to-end. The core idea is to represent previous user-agent interactions and world knowledge in a structured forma heterogeneous graph -and apply message passing to propagate latent representations of users' affinities to refine utterance embeddings. Using these embeddings, PAIGE can potentially provide different rewrites given the same query for users with different preferences. Our model, trained without any human-annotated data, improves the rewrite retrieval precision of state-of-the-art baselines by 12.5-17.5% while having nearly ten times fewer parameters.

show abstract

Section: Related Workmentioning

confidence: 99%

PAIGE: Personalized Adaptive Interactions Graph Encoder for Query Rewriting in Dialogue Systems

Biś,

Gupta,

Hao

et al. 2022

Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing: Industry Track

View full text Add to dashboard Cite

show abstract

“…The current research on mondegreen mainly focuses on pedagogy and linguistics. In addition, some studies are aimed at solving the negative impact of the mondegreen problem on ASR [3], and the research on the generation of the mondegreen itself is blank. Moreover, the research of misheard lyrics generation is insufficient especially in Chinese.…”

Section: Introductionmentioning

confidence: 99%

PreSoramimiset: Establishing dataset for Chinese misheard lyrics generation

2023

ACE

View full text Add to dashboard Cite

Mondegreen is a common phenomenon during conversation and is more obvious during listening to songs. To simulate the impression that the audience will have when they hear a particular piece, the Soramiminet model is introduced in this paper. The model is a combination of wave2vec, and transformer used for generating misheard lyrics in Chinese with various input songs. This article will focus more on the establishment of the dataset for wave2vec model. The criteria and several methods of creating a high-quality dataset are summarized and present in this paper. Additionally, the negative impact of a defective dataset and how to avoid it is discussed. The main limitations and biases of this dataset, and how to address them and future work are explored. The dataset includes a set of audio clips from 21 tracks by 9 different singers in AVI form and a text file for annotation which can be processed by python dataset module. The dataset is relatively biased, since all the annotations are done by the author personally. There is no correct labelling, given that the model is generating misheard or wrong results.

show abstract

On the Turkish Mondegreens and the Reasons for Their Occurrence

Çelik

2022

Korkut Ata Türkiyat Araştırmaları Dergisi

View full text Add to dashboard Cite

“Mondegreen”ler, bir ifadenin veya şarkı sözünün yanlış işitilmesi veya yanlış yorumlanmasından kaynaklanan sözcük dizileridir. Türkçe dışındaki birçok dildeki “mondegreen”ler üzerine muhtelif dil araştırmaları yapılmış ve bunların fonetik ya da fonolojik özellikleri ortaya konmuştur. Bu çalışmada ise Türkçe “mondegreen”ler dilsel açıdan ele alınmış ve bunların ortaya çıkışındaki fonetik ya da fonolojik nedenler Türk dili açısından değerlendirilmeye çalışılmıştır. Bu yapılırken “mondegreen”ler ile ilgili literatür özetine değinilmiş, çalışmanın yöntem ve hedefleri açıklanmış ve Türkçeden seçilen çeşitli “mondegreen” örnekleri üzerinden fonetik ya da fonolojik birçok yorum ortaya konmuştur. Böylelikle, Türkçe “mondegreen”lerin ortaya çıkma sebepleri ve dil özellikleri dikkatlere sunulmuş, sağlıklı bir iletişimin sağlanması yolunda bir engel olarak değerlendirilebilecek yanlış işitmelerin Türk dilindeki yapısı hakkında çeşitli tespitler dile getirilmiştir. Türkçe “mondegreen”lerin dil yapılarını ve meydana çıkış nedenlerini ortaya koymak, diğer dillerdeki “mondegreen” yapıları ile Türk dilindeki “mondegreen” yapılarının dilsel anlamda karşılaştırılmasına olanak sağlayacak, böylece yanlış işitmelerin çeşitli dillerdeki karakteristik nitelikleri hakkında ortak birtakım çıkarımlar yapılabilmesine ortam hazırlayarak sağlıklı iletişimin unsurlarını anlama yolunda bir katkı sağlamış olacaktır.

show abstract

Mondegreen: A Post-Processing Solution to Speech Recognition Error Correction for Voice Search Queries

Cited by 3 publications

References 17 publications

PAIGE: Personalized Adaptive Interactions Graph Encoder for Query Rewriting in Dialogue Systems

PAIGE: Personalized Adaptive Interactions Graph Encoder for Query Rewriting in Dialogue Systems

PreSoramimiset: Establishing dataset for Chinese misheard lyrics generation

On the Turkish Mondegreens and the Reasons for Their Occurrence

Contact Info

Product

Resources

About