CSTR - Downloads
- Corpora, data sets and synthetic voices
- fda database - for evaluating pitch determination algorithms
- SVitchboard 1 - small vocabulary tasks from Switchboard 1
- MOCHA-TIMIT - acoustic + articulatory recordings
- Eustace - corpus for investigating durational effects in speech
- mngu0 - corpus of multimodal articulatory data for one British English male speaker
- REHASP 0.5 - single-speaker corpus of repeated Harvard sentence prompts
- Romanian speech synthesis (RSS) corpus
- NAM corpus - recordings of non-audible murmured speech
- VCTK corpus - from the Voice Clone Toolkit, comprising 109 native speakers of English
- Southern British Female Voice Package - commercial voice package for use with Festival/HTS
- Trained models for ASR or TTS
- Articulatory feature classifier trained on 2000 hours of telephone speech
- GlobalPhone MLPs - phoneme and articulatory feature classifiers trained on several languages
- Pre-built speech synthesis voices for HTS
- Software, tools and pronunciation dictionaries
- Outputs from collaborative projects
- The AMI Corpus
- Blizzard Challenge data and tools
- Hurricane Challenge data
- EMIME - cross-lingual speaker-adaptive speech synthesis
- SSPNet resources for social signal processing
- Simple4All - building speech synthesis systems with very limited resources