Comparison Of Decoding Strategies For CTC Acoustic Models

Abstract

Connectionist Temporal Classification has recently attracted a lot of interest as it offers an elegant approach to building acoustic models (AMs) for speech recognition. The CTC loss function maps an input sequence of observable feature vectors to an output sequence of symbols. Output symbols are conditionally independent of each other under CTC loss, so a language model (LM) can be incorporated conveniently during decoding, retaining the traditional separation of acoustic and linguistic components in ASR. For fixed vocabularies, Weighted Finite State Transducers provide a strong baseline for efficient integration of CTC AMs with n-gram LMs. Character-based neural LMs provide a straight forward solution for open vocabulary speech recognition and all-neural models, and can be decoded with beam search. Finally, sequence-to-sequence models can be used to translate a sequence of individual sounds into a word string. We compare the performance of these three approaches, and analyze their er

Stats

• citations24
• S2 citations—
• github stars0
• HF likes0
• heat score10.48
• arxiv keyzenkel2017comparison

Related papers

• A Study Of All-convolutional Encoders For Connectionist Temporal Classification (2017)5.84
• Blank Collapse: Compressing CTC Emission For The Faster Decoding (2022)0.00
• Back From The Future: Bidirectional CTC Decoding Using Future Information In Speech Recognition (2021)0.00
• CR-CTC: Consistency Regularization On CTC For Improved Speech Recognition (2024)6.30
• Multilingual Training And Cross-lingual Adaptation On Ctc-based Acoustic Model (2017)0.00
• 4D ASR: Joint Modeling Of CTC, Attention, Transducer, And Mask-predict Decoders (2022)7.50
• Adding Connectionist Temporal Summarization Into Conformer To Improve Its Decoder Efficiency For Speech Recognition (2022)0.00
• HMM Vs. CTC For Automatic Speech Recognition: Comparison Based On Full-sum Training From Scratch (2022)0.00