FastSpeech: Fast, Robust and Controllable Text to Speech
Abstract： 提出一种基于transformer的新型前馈网络FastSpeech，和自回归的Transformer TTS相比 将梅尔谱的生成速度提高了近270倍。端到端的语音合成速度提高了38倍，大gpu上语音合成速度达到实时语音速度的30倍 Introduction： 基于神经网络的tts和传统语音合成的拼接法和参数法相比，end to end的TTS通常具有更好的声音自然度。 但在基
相关文章
- 1. 语音合成 | FastSpeech:Fast,Robust and Controllable Text to Speech论文阅读
- 2. Almost Unsupervised Text to Speech and Automatic Speech Recognition
- 3. Voice Conversion by Cascading Automatic Speech Recognition and Text-to-Speech Synthesis with Prosody
- 4. Azure Cognitive Services- Speech To Text
- 5. 5 Open Source Speech Recognition/Speech-to-Text Systems
- 6. 谷歌Text-to-Speech普遍可用，同时发布Speech-to-Text更新
- 7. Hierarchical Generative Modeling for Controllable Speech Synthesis笔记
- 8. Fused DNN: A deep neural network fusion approach to fast and robust pedestrian detect
- 9. 平安-Meta-learning for Multilingual Text-to-Speech
- 10. Windows phone 应用开发[1]-Text To Speech
- 更多相关文章...
- • SVG
- SVG 教程 - • XSL-FO table-and-caption 对象 - XSL-FO 教程
- • RxJava操作符（七）Conditional and Boolean
- • 为了进字节跳动，我精选了29道Java经典算法题，带详细讲解
相关标签/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
欢迎关注本站公众号,获取更多信息
相关文章
- 1. 语音合成 | FastSpeech:Fast,Robust and Controllable Text to Speech论文阅读
- 2. Almost Unsupervised Text to Speech and Automatic Speech Recognition
- 3. Voice Conversion by Cascading Automatic Speech Recognition and Text-to-Speech Synthesis with Prosody
- 4. Azure Cognitive Services- Speech To Text
- 5. 5 Open Source Speech Recognition/Speech-to-Text Systems
- 6. 谷歌Text-to-Speech普遍可用，同时发布Speech-to-Text更新
- 7. Hierarchical Generative Modeling for Controllable Speech Synthesis笔记
- 8. Fused DNN: A deep neural network fusion approach to fast and robust pedestrian detect
- 9. 平安-Meta-learning for Multilingual Text-to-Speech
- 10. Windows phone 应用开发[1]-Text To Speech