本页面单独归档日常积累的技术分享和 Talk Slides，涵盖语音合成、语音识别、音频生成等多个方向，方便查阅。
2025 年
2025-11-03 · Recent Advances in Autoregressive Diffusion Models for Speech Generation语音合成 扩散模型+自回归
关键工作：MELA-TTS, VoxCPM, VibeVoice, Ming-UniAudio
2025-08-21 · DiffRO: Differentiable Reward Optimization for LLM based TTS system语音合成 强化学习
关键工作：CosyVoice, CosyVoice2
2025-06-26 · Rethinking Tortoise TTS Paradigm语音合成 TTS 大模型
关键工作：Tortoise-TTS, XTTS, Index-TTS
2025-04-14 · MegaTTS 3: Sparse Alignment Enhanced Latent Diffusion Transformer for Zero-Shot Speech Synthesis语音合成 扩散模型
关键工作：MegaTTS 3, InspireMusic, RALL-E, VALLE-R, VoiceLDM, PeRFlow
2024 年
2024-09-25 · FireRedTTS: A Foundation Text-To-Speech Framework for Industry-Level Generative Speech Applications语音合成 扩散模型
关键工作：FireRedTTS, Semantic-Aware Speech Tokenizer
2024-09-23 · Moshi: A Speech-Text Foundation Model for Real-Time Dialogue语音对话 全双工大模型
关键工作：Helium, Mimi, Moshi
2024-07-09 · CosyVoice: A Scalable Multilingual Zero-shot Text-to-speech Synthesizer based on Supervised Semantic Tokens语音合成 大模型
关键工作：CosyVoice
2024-04-01 · NaturalSpeech 3: Zero-Shot Speech Synthesis with Factorized Codec and Diffusion Models语音合成 离散 Diffusion
关键工作：FACodec, NaturalSpeech 3, MaskGIT, Discrete Diffusion
2024-03-06 · MobileSpeech: A Fast and High-Fidelity Framework for Mobile Zero-Shot Text-to-Speech端侧语音合成 高效推理
关键工作：MobileSpeech, SoundStorm, Pheme
2023 年
2023-11-20 · Recent Advances on Unified Model for Voice and Audio Generation (Part 2)语音与音频生成 扩散模型
关键工作：AudioLDM 2, Stable Diffusion
2023-11-13 · Recent Advances on Unified Model for Voice and Audio Generation (Part 1)语音与音频生成 LLM 模型
关键工作：Speech-X, Make-A-Voice / MVoice, UniAudio
2023-06-01 · Speak, Read and Prompt High-Fidelity Text-to-Speech with Minimal Supervision语音合成
关键工作：w2v-BERT, SoundStream, Spear-TTS
2023-01-12 · Text-Based Speech Editing From a Perspective of Mask-Based Generative Models语音编辑 语音合成
关键工作：SpeechPainter, A3T, MUSE (文生图)
2022 年
2022-11-13 · Voice Cloning C: Advanced Methods声音复刻 Perceiver Adapter
关键工作：RetrieverTTS, TTS Adapters
2022-09-15 · Voice Cloning B: Overview on Transfer Learning, Representation Learning and Meta-Learning声音复刻 迁移学习 元学习
关键工作：ProsodySpeech, StyleSpeech, GST, Meta-TTS
2022-05-26 · Voice Cloning A: Recent Advances in Adaptive Text-to-Speech声音复刻 自适应TTS 少样本
关键工作：AdaSpeech, AdaSpeech2, AdaSpeech3, AdaSpeech4
2021 年
2021-07-01 · Lattice: Concepts, Methods and ApplicationsLattice 语言模型 WFST
关键工作：WFST, Lattice, Lattice-Transformer
2020 年
2020-09-17 · An Overview of Hybrid ASR: Acoustic Models, Language Models, and Post-Processing语音识别 声学模型 语言模型
关键工作：LF-MMI, TDNN-F, Lattice Rescoring
2020-07-20 · Personal Work on Speech Recognition (Bachelor/Master)语音识别 说话人自适应 迁移学习
关键工作：Speaker Adaption, Transfer Learning