AI Audio Datasets (AI-ADS) aggregates publicly available audio corpora across speech, music, and sound effect categories for AI training purposes. It catalogs resources like AISHELL for Mandarin speech recognition, AISHELL-3 for multi-speaker TTS, and Audio-FLAN for unified audio-language model instruction-tuning. The repository serves as a reference index for researchers and developers seeking training data for audio generation, speech synthesis, and audio understanding models.