AI audio enhancer

musicgen-songstarter-v0.2 is an audio generation model designed for music producers, specifically used to generate useful melody loops. The model was fine-tuned on the melody loop dataset from the Splice sample library and was able to generate stereo audio with an audio frequency of 32kHz. Compared to v0.1, v0.2 uses three times more unique samples and increases the model size from medium to large.

stable-audio-tools is an open source PyTorch library that provides training and inference code for generative models for conditional audio generation. Including autoencoders, implicit diffusion models, MusicGen, etc. Supports multi-GPU training and can generate high-quality audio.

Ultimate Vocal Removal GUI is a vocal removal tool using deep neural network technology. Its core developers trained all provided models except Demucs v3 and v4 4-channel models. The application uses advanced source separation models to remove vocals from audio files. No additional prerequisites are required to run effectively. Available for Windows 10 and above.

NVAS3d is a project for estimating sound at any location in a scene containing multiple unknown sound sources, enabling a new perspective on acoustic synthesis by using audio recordings from multiple microphones and the 3D geometry and materials of the scene.

AudioSep is an open-domain audio source separation model based on natural language query. It consists of two key components: text encoder and separation model. We train AudioSep on a large-scale multi-modal dataset and extensively evaluate its capabilities on a number of tasks, including audio event separation, instrument separation, and speech enhancement. AudioSep demonstrates strong separation performance and impressive zero-shot generalization capabilities using audio titles or text labels as queries, significantly outperforming previous sound separation models for audio queries and language queries. To ensure the reproducibility of this work, we will release the source code, evaluation benchmarks, and pretrained models.

Enhance Speech from Adobe is a free AI audio filter that makes spoken audio look like it was recorded in a soundproof studio. It can automatically remove background noise, adjust volume balance, and improve audio quality. Users can upload recording files to the platform, and the audio will be optimized through AI algorithms. Enhance Speech from Adobe is suitable for broadcasting, podcasting, audio production and other fields. This product is completely free to use.