Hoon-Young Cho

Hanbin Bae, Pavel Andreev, Azat Saginbaev et al.

This paper introduces a speech enhancement solution tailored for true wireless stereo (TWS) earbuds on-device usage. The solution was specifically designed to support conversations in noisy environments, with active noise cancellation (ANC) activated. The primary challenges for speech enhancement models in this context arise from computational complexity that limits on-device usage and latency that must be less than 3 ms to preserve a live conversation. To address these issues, we evaluated several crucial design elements, including the network architecture and domain, design of loss functions, pruning method, and hardware-specific optimization. Consequently, we demonstrated substantial improvements in speech enhancement quality compared with that in baseline models, while simultaneously reducing the computational complexity and algorithmic latency.

Youngmoon Jung, Myunghun Jung, Joon-Young Yang et al.

Open-vocabulary keyword spotting (KWS) with text-based enrollment has emerged as a flexible alternative to fixed-phrase triggers. Prior utterance-level matching methods, from an embedding-learning standpoint, learn embeddings at a single fixed dimensionality. We depart from this design and propose Matryoshka Audio-Text Embeddings (MATE), a dual-encoder framework that encodes multiple embedding granularities within a single vector via nested sub-embeddings ("prefixes"). Specifically, we introduce a PCA-guided prefix alignment: PCA-compressed versions of the full text embedding for each prefix size serve as teacher targets to align both audio and text prefixes. This alignment concentrates salient keyword cues in lower-dimensional prefixes, while higher dimensions add detail. MATE is trained with standard deep metric learning objectives for audio-text KWS, and is loss-agnostic. To our knowledge, this is the first application of matryoshka-style embeddings to KWS, achieving state-of-the-art results on WSJ and LibriPhrase without any inference overhead.

Youngmoon Jung, Joon-Young Yang, Ju-ho Kim et al.

Short-utterance speaker verification remains challenging due to limited speaker-discriminative cues in short speech segments. While existing methods focus on enhancing speaker encoders, the embedding learning strategy still forces a single fixed-dimensional representation reused for utterances of any length, leaving capacity misaligned with the information available at different durations. We propose Duration-Aware Matryoshka Embedding (DAME), a model-agnostic framework that builds a nested hierarchy of sub-embeddings aligned to utterance durations: lower-dimensional representations capture compact speaker traits from short utterances, while higher dimensions encode richer details from longer speech. DAME supports both training from scratch and fine-tuning, and serves as a direct alternative to conventional large-margin fine-tuning, consistently improving performance across durations. On the VoxCeleb1-O/E/H and VOiCES evaluation sets, DAME consistently reduces the equal error rate on 1-s and other short-duration trials, while maintaining full-length performance with no additional inference cost. These gains generalize across various speaker encoder architectures under both general training and fine-tuning setups.

Youngmoon Jung, Jinyoung Lee, Seungjin Lee et al.

Recent advances in flexible keyword spotting (KWS) with text enrollment allow users to personalize keywords without uttering them during enrollment. However, there is still room for improvement in target keyword performance. In this work, we propose a novel few-shot transfer learning method, called text-aware adapter (TA-adapter), designed to enhance a pre-trained flexible KWS model for specific keywords with limited speech samples. To adapt the acoustic encoder, we leverage a jointly pre-trained text encoder to generate a text embedding that acts as a representative vector for the keyword. By fine-tuning only a small portion of the network while keeping the core components' weights intact, the TA-adapter proves highly efficient for few-shot KWS, enabling a seamless return to the original pre-trained model. In our experiments, the TA-adapter demonstrated significant performance improvements across 35 distinct keywords from the Google Speech Commands V2 dataset, with only a 0.14% increase in the total number of parameters.

Youngmoon Jung, Yong-Hyeok Lee, Myunghun Jung et al.

For text enrollment-based open-vocabulary keyword spotting (KWS), acoustic and text embeddings are typically compared at either the phoneme or utterance level. To facilitate this, we optimize acoustic and text encoders using deep metric learning (DML), enabling direct comparison of multi-modal embeddings in a shared embedding space. However, the inherent heterogeneity between audio and text modalities presents a significant challenge. To address this, we propose Modality Adversarial Learning (MAL), which reduces the domain gap in heterogeneous modality representations. Specifically, we train a modality classifier adversarially to encourage both encoders to generate modality-invariant embeddings. Additionally, we apply DML to achieve phoneme-level alignment between audio and text, and conduct extensive comparisons across various DML objectives. Experiments on the Wall Street Journal (WSJ) and LibriPhrase datasets demonstrate the effectiveness of the proposed approach.

Hanbin Bae, Byungjun Kang, Jiwon Kim et al.

This study emphasizes the significance of exploring distance-based source separation (DSS) in outdoor environments. Unlike existing studies that primarily focus on indoor settings, the proposed model is designed to capture the unique characteristics of outdoor audio sources. It incorporates advanced techniques, including a two-stage conformer block, a linear relation-aware self-attention (RSA), and a TensorFlow Lite GPU delegate. While the linear RSA may not capture physical cues as explicitly as the quadratic RSA, the linear RSA enhances the model's context awareness, leading to improved performance on the DSS that requires an understanding of physical cues in outdoor and indoor environments. The experimental results demonstrated that the proposed model overcomes the limitations of existing approaches and considerably enhances energy efficiency and real-time inference speed on mobile devices.

Youngmoon Jung, Seungjin Lee, Joon-Young Yang et al.

In recent years, there has been an increasing focus on user convenience, leading to increased interest in text-based keyword enrollment systems for keyword spotting (KWS). Since the system utilizes text input during the enrollment phase and audio input during actual usage, we call this task audio-text based KWS. To enable this task, both acoustic and text encoders are typically trained using deep metric learning loss functions, such as triplet- and proxy-based losses. This study aims to improve existing methods by leveraging the structural relations within acoustic embeddings and within text embeddings. Unlike previous studies that only compare acoustic and text embeddings on a point-to-point basis, our approach focuses on the relational structures within the embedding space by introducing the concept of Relational Proxy Loss (RPL). By incorporating RPL, we demonstrated improved performance on the Wall Street Journal (WSJ) corpus.

Gyeong-Hoon Lee, Tae-Woo Kim, Hanbin Bae et al.

Recently, end-to-end Korean singing voice systems have been designed to generate realistic singing voices. However, these systems still suffer from a lack of robustness in terms of pronunciation accuracy. In this paper, we propose N-Singer, a non-autoregressive Korean singing voice system, to synthesize accurate and pronounced Korean singing voices in parallel. N-Singer consists of a Transformer-based mel-generator, a convolutional network-based postnet, and voicing-aware discriminators. It can contribute in the following ways. First, for accurate pronunciation, N-Singer separately models linguistic and pitch information without other acoustic features. Second, to achieve improved mel-spectrograms, N-Singer uses a combination of Transformer-based modules and convolutional network-based modules. Third, in adversarial training, voicing-aware conditional discriminators are used to capture the harmonic features of voiced segments and noise components of unvoiced segments. The experimental results prove that N-Singer can synthesize a natural singing voice in parallel with a more accurate pronunciation than the baseline model.

Jinhyeok Yang, Jae-Sung Bae, Taejun Bak et al.

Recent advances in neural multi-speaker text-to-speech (TTS) models have enabled the generation of reasonably good speech quality with a single model and made it possible to synthesize the speech of a speaker with limited training data. Fine-tuning to the target speaker data with the multi-speaker model can achieve better quality, however, there still exists a gap compared to the real speech sample and the model depends on the speaker. In this work, we propose GANSpeech, which is a high-fidelity multi-speaker TTS model that adopts the adversarial training method to a non-autoregressive multi-speaker TTS model. In addition, we propose simple but efficient automatic scaling methods for feature matching loss used in adversarial training. In the subjective listening tests, GANSpeech significantly outperformed the baseline multi-speaker FastSpeech and FastSpeech2 models, and showed a better MOS score than the speaker-specific fine-tuned FastSpeech2.

Taejun Bak, Jae-Sung Bae, Hanbin Bae et al.

Methods for modeling and controlling prosody with acoustic features have been proposed for neural text-to-speech (TTS) models. Prosodic speech can be generated by conditioning acoustic features. However, synthesized speech with a large pitch-shift scale suffers from audio quality degradation, and speaker characteristics deformation. To address this problem, we propose a feed-forward Transformer based TTS model that is designed based on the source-filter theory. This model, called FastPitchFormant, has a unique structure that handles text and acoustic features in parallel. With modeling each feature separately, the tendency that the model learns the relationship between two features can be mitigated.

Hanbin Bae, Jae-Sung Bae, Young-Sun Joo et al.

Recently, it has become easier to obtain speech data from various media such as the internet or YouTube, but directly utilizing them to train a neural text-to-speech (TTS) model is difficult. The proportion of clean speech is insufficient and the remainder includes background music. Even with the global style token (GST). Therefore, we propose the following method to successfully train an end-to-end TTS model with limited broadcast data. First, the background music is removed from the speech by introducing a music filter. Second, the GST-TTS model with an auxiliary quality classifier is trained with the filtered speech and a small amount of clean speech. In particular, the quality classifier makes the embedding vector of the GST layer focus on representing the speech quality (filtered or clean) of the input speech. The experimental results verified that the proposed method synthesized much more high-quality speech than conventional methods.

Jae-Sung Bae, Hanbin Bae, Young-Sun Joo et al.

This paper proposes a controllable end-to-end text-to-speech (TTS) system to control the speaking speed (speed-controllable TTS; SCTTS) of synthesized speech with sentence-level speaking-rate value as an additional input. The speaking-rate value, the ratio of the number of input phonemes to the length of input speech, is adopted in the proposed system to control the speaking speed. Furthermore, the proposed SCTTS system can control the speaking speed while retaining other speech attributes, such as the pitch, by adopting the global style token-based style encoder. The proposed SCTTS does not require any additional well-trained model or an external speech database to extract phoneme-level duration information and can be trained in an end-to-end manner. In addition, our listening tests on fast-, normal-, and slow-speed speech showed that the SCTTS can generate more natural speech than other phoneme duration control approaches which increase or decrease duration at the same rate for the entire sentence, especially in the case of slow-speed speech.

Yoonjae Jeong, Hoon-Young Cho

The purpose of this study is to detect the mismatch between text script and voice-over. For this, we present a novel utterance verification (UV) method, which calculates the degree of correspondence between a voice-over and the phoneme sequence of a script. We found that the phoneme recognition probabilities of exaggerated voice-overs decrease compared to ordinary utterances, but their rankings do not demonstrate any significant change. The proposed method, therefore, uses the recognition ranking of each phoneme segment corresponding to a phoneme sequence for measuring the confidence of a voice-over utterance for its corresponding script. The experimental results show that the proposed UV method outperforms a state-of-the-art approach using cross modal attention used for detecting mismatch between speech and transcription.