There are several opensource AI speech projects available、Here are a few notable ones:

1、Mozilla DeepSpeech: This is an opensource speechtotext engine developed by Mozilla、It's based on a deep learning approach and can be trained on various datasets.

2、CMU Sphinx: Carnegie Mellon University's Sphinx is another popular opensource speech recognition system、It's known for its flexibility and has been used in many research projects.

3、Kaldi: Developed at Johns Hopkins University, Kaldi is a toolkit for building speech recognition systems、It's widely used in both academia and industry.

4、OpenEars: This is an iOS library for speech recognition that uses the iPhone's builtin speech recognition capabilities.

5、pocketsphinx: A smaller, faster version of CMU Sphinx designed for embedded systems.

6、Julius: An opensource large vocabulary continuous speech recognition engine originally developed at ATR Spoken Language Communication Research Laboratories in Japan.

7、Wav2Letter++: Developed by Facebook AI Research, this is a fast automatic speech recognition engine based on convolutional neural networks.

8、Coqui STT (SpeechToText): An opensource, multilingual automatic speech recognition engine.

These projects vary in terms of complexity, supported languages, and ease of use、Depending on your specific needs, one of these might be more suitable than others.