SafeEar is an innovative audio depth detection framework that can identify deep audio without relying on speech content. The framework employs a neural audio codec design to separate semantic and acoustic information from audio samples, using solely acoustic features (such as prosody and timbre) for depth detection, thereby safeguarding the privacy of speech content. SafeEar enhances the detector's capabilities by augmenting the codec in real-world scenarios, enabling it to recognize various forms of deep audio. Extensive experiments conducted on four benchmark datasets demonstrate that SafeEar is highly effective in detecting various deep technologies, achieving a minimum equal error rate (EER) of 2.02%. Moreover, it protects speech content in five languages from deciphering by both machine and human auditory analysis, as validated by our user studies revealing a word error rate (WER) exceeding 93.93%. Additionally, SafeEar has established a benchmark for anti-deep and anti-content recovery assessments, laying the groundwork for future research in audio privacy protection and depth detection.

Stable Audio ControlNet is a music generation model based on Stable Audio Open, fine-tuned with DiT ControlNet. It can operate on GPUs with 16GB VRAM and supports audio control. Although still in development, it is capable of generating and controlling music, offering significant technical implications and application potential.

Stable Audio Open 1.0 is an AI model that utilizes an autoencoder, T5-based text embeddings, and a transformer-based diffusion model to generate up to 47 seconds of stereo audio. It generates music and audio through text prompts, supporting research and experiments to explore the current capabilities of generative AI models. The model is trained on datasets from Freesound and the Free Music Archive (FMA), ensuring data diversity and copyright legality.

Make-An-Audio 2 is a text-to-audio generation technology based on diffusion models, co-developed by researchers from Zhejiang University, ByteDance, and the Chinese University of Hong Kong. This technology utilizes pre-trained large language models (LLMs) to parse text, optimizing for semantic alignment and temporal consistency, thereby improving the quality of generated audio. It also incorporates a feed-forward Transformer-based diffusion denoiser to enhance performance in generating variable-length audio and bolster the extraction of temporal information. Furthermore, by leveraging LLMs to convert abundant audio label data into audio-text datasets, the issue of time data scarcity is addressed.

Suno API is a non-official Suno AI client based on Python and Streamlit, currently supporting music generation and retrieval of music information functionalities. It comes with built-in maintenance and activation features, removing concerns about token expiration. Users can set multiple accounts for storage and usage. Additionally, it supports multiple languages, including Chinese, English, Korean, Japanese, etc.

Suno Music Generator is an unofficial NextJS-based suno.ai music generation website. Users can generate desired songs by inputting a prompt. Through JavaScript reverse engineering and parsing of suno.ai's song generation API, it includes built-in token refresh and keep-alive functionality, eliminating worries about token expiration.

musicgen-songstarter-v0.2 is an audio generation model designed for music producers, specializing in generating useful melody loops. Fine-tuned on a melody loop dataset from the Splice sample library, it can generate stereo audio with a frequency of 32kHz. Compared to version v0.1, v0.2 utilizes three times the unique samples and the model size has increased from medium to large.

SunoAPI is an unofficial Suno API built on Python and FastAPI. It supports features like generating songs and lyrics, and includes built-in token management and activity preservation, so you don't have to worry about token expiration. SunoAPI utilizes an entirely asynchronous design, resulting in fast operation and suitability for future expansion. Users can easily generate various music content using the API.

ChatMusician is an open-source Large Language Model (LLM) that integrates musical capabilities through continual pre-training and fine-tuning. The model is based on text-compatible musical notation (ABC notation) and treats music as a second language. ChatMusician can understand and generate music without relying on external multimodal neural structures or tokenizer.

Remusic is an AI-driven music creation platform that allows users to generate unique music, lyrics, and album covers through simple inputs. This platform transforms traditional music creation methods, enabling those without a professional music background to easily produce high-quality music works. Remusic's main advantages include ease of use, diversity, and innovation, supporting a variety of music styles and genres while providing royalty-free music suitable for both personal and commercial use. Background information on Remusic indicates it has over 2,000 creators and more than 10,000 pieces of music and songs, with a user satisfaction rate of 98%.

SafeEar is an innovative audio depth detection framework that can identify deep audio without relying on speech content. The framework employs a neural audio codec design to separate semantic and acoustic information from audio samples, using solely acoustic features (such as prosody and timbre) for depth detection, thereby safeguarding the privacy of speech content. SafeEar enhances the detector's capabilities by augmenting the codec in real-world scenarios, enabling it to recognize various forms of deep audio. Extensive experiments conducted on four benchmark datasets demonstrate that SafeEar is highly effective in detecting various deep technologies, achieving a minimum equal error rate (EER) of 2.02%. Moreover, it protects speech content in five languages from deciphering by both machine and human auditory analysis, as validated by our user studies revealing a word error rate (WER) exceeding 93.93%. Additionally, SafeEar has established a benchmark for anti-deep and anti-content recovery assessments, laying the groundwork for future research in audio privacy protection and depth detection.

Covers is a music creation tool launched by Suno that allows users to convert any sound, from simple voice memos to fully produced tracks, into new musical works while retaining the original melody. This feature is particularly suitable for music creators wanting to explore different styles or add lyrics to instrumental tracks. Currently, Covers is in early access testing, available to all Pro and Premier subscribers, allowing users to try it for free up to 200 times. After that, each new cover generated will consume 10 credits.

Stable Audio ControlNet is a music generation model based on Stable Audio Open, fine-tuned with DiT ControlNet. It can operate on GPUs with 16GB VRAM and supports audio control. Although still in development, it is capable of generating and controlling music, offering significant technical implications and application potential.

SAAR is an AI music assistant designed to help users create music through intuitive voice or text conversations. It is similar to Siri or Alexa, but focused specifically on music creation. SAAR can generate music and lyrics, facilitating musical discussions with users. It is crafted for musicians and dreamers, aiming to simplify the music creation process and make creativity easier to achieve.

Udio v1.5 is an upgraded version of a music creation platform with multiple enhancements over version 1, including improved audio quality, pitch control, and better global language support. It generates 48kHz stereo tracks, offering clearer sound quality and better instrument separation. Additionally, Udio v1.5 features a range of new capabilities such as a dedicated creation page, track downloads, audio-to-audio mixing, and shareable lyric videos, all aimed at empowering music creators further.

Stable Audio Open 1.0 is an AI model that utilizes an autoencoder, T5-based text embeddings, and a transformer-based diffusion model to generate up to 47 seconds of stereo audio. It generates music and audio through text prompts, supporting research and experiments to explore the current capabilities of generative AI models. The model is trained on datasets from Freesound and the Free Music Archive (FMA), ensuring data diversity and copyright legality.

Trip Tunes is an app designed for travelers that automatically creates music playlists tailored to the atmosphere of their trip. Users simply enter basic travel information and their music tastes, and the app uses algorithms to curate tracks that match local music scenes and travel styles. This app is especially suitable for those who want to enjoy personalized music experiences while traveling.

Make-An-Audio 2 is a text-to-audio generation technology based on diffusion models, co-developed by researchers from Zhejiang University, ByteDance, and the Chinese University of Hong Kong. This technology utilizes pre-trained large language models (LLMs) to parse text, optimizing for semantic alignment and temporal consistency, thereby improving the quality of generated audio. It also incorporates a feed-forward Transformer-based diffusion denoiser to enhance performance in generating variable-length audio and bolster the extraction of temporal information. Furthermore, by leveraging LLMs to convert abundant audio label data into audio-text datasets, the issue of time data scarcity is addressed.