Latest Articles for Speech Recognition

TOGGL model improves transcription accuracy for overlapping speech situations.

A method to enhance speech recognition quality in noisy environments.

Researchers develop SaSLaW to enhance machine speech adaptation in various environments.

A new dataset highlights biases in speech models based on gender and age.

Research reveals how to make speech models smaller and more efficient.

Adversarial training enhances keyword spotting accuracy in synthetic and real speech.

A new benchmark improves evaluation of speech emotion recognition systems across languages and emotions.

New methods enhance ASR models for multiple languages, preserving past knowledge.

A new approach enhances recognition of code-switched phrases in bilingual speech.

A new method for better handling of long data sequences.

Examining how voice patterns affect meaning and technology performance.

A look into the complexities of identifying mixed audio tracks.

O-HuBERT enhances speech recognition by separating content and expressive information.

A new method improves speech recognition for Hindi using pseudo-labeling techniques.

A system to classify Literary and Colloquial Tamil dialects using sound features.

New methods enhance computer understanding of whispered and normal speech.

A look at micro-batch clipping and its benefits for model training.

Research shows how LLMs enhance automatic speech recognition in Japanese language.

This article examines how models recognize tone, stress, and pitch accents.

SALSA enhances speech recognition accuracy for low-resource languages by integrating ASR and language models.

New method enhances ASR accuracy using language models for better transcriptions.

A new system corrects speaker identification errors for clearer conversation transcripts.

Improving speech clarity through hybrid filterbanks and neural networks.

A new model enhances speech recognition by combining audio and visual inputs effectively.

New methods improve speech recognition in challenging multi-speaker situations.

A new method improves automatic speech recognition by preserving sound order in knowledge transfer.

This study examines how noise can enhance speech recognition resilience against challenges.

Innovative lightweight transducer enhances speech recognition efficiency and accuracy.

This article compares discrete and continuous speech representations for effective speech recognition.

Research reveals how neurons in speech models recognize key features of sound.

This study examines how self-attention affects speech recognition in Turkish and English.

A self-supervised learning approach reduces the need for labeled audio data.

TF-Mamba enhances sound localization using a novel approach integrating time and frequency data.

Research on modular ASR systems aims to improve performance in noisy environments.

Introducing DENSE, a method enhancing target speech extraction using dynamic embeddings.

This method enhances recognition accuracy for uncommon names in speech outputs.

Enhancing spoken word identification through visual cues in under-resourced languages.

BigCodec improves sound quality in low-bitrate audio transmission.

This article discusses the benefits of simplifying transformer models for speech tasks.

Sortformer integrates speaker diarization and ASR for improved audio processing.