Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 1709.07552

Techniques and Challenges in Speech Synthesis

David Ferris
View original source ↗Read source document ↗Export metadata ↓

Abstract

The aim of this project was to develop and implement an English language Text-to-Speech synthesis system. This involved a study of mechanisms of human speech production, a review of techniques in speech synthesis, and analysis of tests used to evaluate the effectiveness of synthesized speech. It was determined that a diphone synthesis system was the most effective choice for the scope of this project. A method of automatically identifying and extracting diphones from prompted speech was designed, allowing for the creation of a diphone database by a speaker in less than 40 minutes. CMUdict was used to determine the pronunciation of known words. A system for smoothing the transitions between diphone recordings was designed and implemented. CMUdict was then used to train a maximum-likelihood prediction system to determine the correct pronunciation of unknown English language alphabetic words. Then, a Part Of Speech tagger was designed to find the lexical class of words within a sentence. A method of altering the pitch, duration, and volume of the produced voice over time was designed, being a combination of the TD-PSOLA algorithm and a novel approach referred to as Unvoiced Speech Duration Shifting. This minimises distortion of the voice when shifting the pitch or duration, while maximising computational efficiency by operating in the time domain. This approach was used to add correct lexical stress to vowels within words. A text tokenisation system was developed to handle arbitrary text input, allowing pronunciation of numerical input tokens and use of appropriate pauses for punctuation. Methods for further improving sentence level speech naturalness were discussed. Finally, the system was tested with listeners for its intelligibility and naturalness.

Explore related subjects

cs.SDeess.AS

Keep this discovery

BibTeXRIS

David Ferris. 2017-09-22. Techniques and Challenges in Speech Synthesis. https://arxiv.org/abs/1709.07552

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Xiaomi-CocktailASR-1 Technical Report

Recently, large language model (LLM) based ASR models have achieved significant progress, yet they generally lack support for multi-speaker scenarios, where the cocktail party problem remains a critical bottleneck for further advancing ASR. Existing TS-ASR methods, including end-to-end architectures with speaker embeddings and latest LLM-based explorations suffer from degraded single-speaker performance and the inability to reject when the target speaker is absent. In this paper, we propose Xiaomi-CocktailASR-1, an LLM-based end-to-end TS-ASR architecture. By utilizing reference speech as voiceprint prompts, it directly transcribes the target speaker's speech without requiring speech separation. Xiaomi-CocktailASR-1 maintains competitive performance in single-speaker scenarios, comparable to mainstream ASR models. It also features a negative sample rejection capability, outputting empty text when the target speaker is absent from the mixed speech. Additionally, Xiaomi-CocktailASR-1 supports a Chain-of-Thought (CoT) reasoning mode to provide explicit reasoning steps. Extensive experiments on various synthetic and real-world multispeaker benchmarks demonstrate that Xiaomi-CocktailASR-1 achieves state-of-the-art performance, effectively addressing the cocktail party problem through a unified architecture that balances multispeaker and single-speaker recognition accuracy, along with rejection capability.

cs.SD↗
arXiv2026

EConv-TasNet: Efficient Conv-TasNet for Effective Speech Separation

Conv-TasNet has served as a strong baseline for time-domain speech separation, and many studies have extended it with advanced architectures such as dual-path networks, U-Nets, and attention mechanisms. However, these methods often introduce high computational cost and complexity, limiting their deployment in resource-constrained scenarios. To address this issue, we propose eConv-TasNet, an efficient variant of Conv-TasNet that improves both effectiveness and efficiency without relying on resource-intensive modules. The proposed model consists of a group-wise early-splitting (GES) module and a multi-group feature aggregation (MGFA) module. GES generates discriminative speaker embeddings at intermediate stages, while MGFA progressively aggregates these group-level representations for refined mask estimation. Experimental results show that eConv-TasNet reduces model size by 22.4%, accelerates inference by 18.9%, and improves SI-SNRi by 14.0%-28.0% across three public benchmarks. Moreover, it achieves competitive performance compared with state-of-the-art methods while requiring significantly fewer parameters and lower inference cost. These results demonstrate a favorable efficiency-effectiveness trade-off for edge deployment.

cs.SD↗