Using Phonemes in cascaded S2S translation pipeline

Rene Pilz, Johannes Schneider
4/22/2025
cs.LGcs.CLcs.SD

No official code found. Best implementation: fungus75/Phonemes_S2S_Pipeline (0 stars, last commit 2025-05-20).

Abstract

This paper explores the idea of using phonemes as a textual representation within a conventional multilingual simultaneous speech-to-speech translation pipeline, as opposed to the traditional reliance on text-based language representations. To investigate this, we trained an open-source sequence-to-sequence model on the WMT17 dataset in two formats: one using standard textual representation and the other employing phonemic representation. The performance of both approaches was assessed using the BLEU metric. Our findings shows that the phonemic approach provides comparable quality but offers several advantages, including lower resource requirements or better suitability for low-resource languages.

View on arXivView PDF

Code Implementations(1)

fungus75/Phonemes_S2S_PipelineCommunity100%

Additional Data for the Paper "Using Phonemes in cascaded S2S translation pipeline"

00Apr 19, 20251 years ago

Similar Papers

Speech-to-Text Translation with Phoneme-Augmented CoT: Enhancing Cross-Lingual Transfer in Low-Resource Scenarios

85% match2025

MSLM-S2ST: A Multitask Speech Language Model for Textless Speech-to-Speech Translation with Speaker Style Preservation

80% match2024

Direct Speech to Speech Translation: A Review

80% match2025

LLM-Powered Grapheme-to-Phoneme Conversion: Benchmark and Case Study

80% match1 repo192024

SeamlessExpressiveLM: Speech Language Model for Expressive Speech-to-Speech Translation with Chain-of-Thought

79% match2024

From Babble to Words: Pre-Training Language Models on Continuous Streams of Phonemes

79% match1 repo2024

Cite this paper

@article{pilz2025using,
  title  = {Using Phonemes in cascaded S2S translation pipeline},
  author = {Rene Pilz and Johannes Schneider},
  year   = {2025},
  eprint = {2504.16234},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2504.16234v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2504.16234.svg)](https://codeofpaper.com/paper/2504.16234)

Discussion