“Audio Language Models” - Neil Zeghidour
Автор: TTIC
Загружено: 2025-09-25
Просмотров: 735
Описание:
“Audio Language Models”
Neil Zeghidour, Kyutai
Originally recorded on September 5, 2025 as part of the TTIC Summer Workshop on Foundations of Speech and Audio Foundation Models.
In this talk, Neil Zeghidour introduces the concept of audio language models, which unify audio analysis and synthesis within a single framework. By discretizing audio signals with neural codecs and framing them as sequence-to-sequence tasks, these models enable powerful new applications in speech modeling, zero-shot voice conversion, text-to-music generation, and real-time spoken dialogue—highlighting the emerging versatility of language-model-inspired architectures for audio.
Timestamps:
00:00 Introduction
01:35:00 Talk begins
2:28:10 Q&A
#AudioLanguageModels #SpeechAI #AI #MachineLearning #ML #NLP #NaturalLanguageProcessing #MultimodalAI #AudioProcessing #DeepLearning #LLMs #TTIC
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке:
