Sorry, you need to enable JavaScript to visit this website.

Investigating End-to-end ASR Architectures for Long form Audio Transcription

DOI:
10.60864/s9sh-5f97
Citation Author(s):
Nithin Rao Koluguri, Samuel Kriman, Georgy Zelenfroind, Somshubra Majumdar, Dima Rekesh, Vahid Noroozi, Jagadeesh Balam, Boris Ginsburg
Submitted by:
Somshubra Majumdar
Last updated:
6 June 2024 - 10:21am
Document Type:
Presentation Slides
Document Year:
2024
Event:
Presenters:
Somshubra Majumdar
Paper Code:
SS-L22.5
 

This paper presents an overview and evaluation of some of the end-to-end ASR models on long-form audios. We study three categories of Automatic Speech Recognition(ASR) models based on their core architecture: (1) convolutional, (2) convolutional with squeeze-and-excitation and (3) convolutional models with attention. We selected one ASR model from each category and evaluated Word Error Rate, maximum audio length and real-time factor for each model on a variety of long audio benchmarks: Earnings-21 and 22, CORAAL, and TED-LIUM3. The model from the category of self-attention with local attention and global token has the best accuracy comparing to other architectures. We also compared models with CTC and RNNT decoders and showed that CTC-based models are more robust and efficient than RNNT on long form audio.

up
0 users have voted: