软餐获悉，英特尔为开源音频编辑程序 Audacity 发布了一个名为 OpenVINO AI 的 AI 工具套件。这些插件允许 Audacity 用户在本地转录播客并根据提示生成音乐。OpenVINO AI 仅与适用于 Windows 的 64 位版 Audacity 3.4.2 兼容，并且必须从 GitHub 提供的文件中手动安装。Linux 或 macOS 上的用户现在需要自行编译。Audacity 开发商称，其中包括三个音乐工具，即音乐生成、音乐风格混音和音乐分离。前两个工具使用 Riffusion 模型（Stable Diffusion 的一个分支）根据文本提示生成或重新混合音乐；音乐分离工具允许用户将现有歌曲拆分为不同的曲目。该工具可以将一首歌分解为人声和乐器，或分解为人声、鼓和贝斯以及其余乐器的四个音轨。
此外，还提供了两种用于语音音频内容的工具：噪声抑制和转录。这些插件允许用户分别使用 AI 去除背景噪音和转录语音音频。所有 OpenVINO AI 工具都在本地运行，因此它们不需要连接到英特尔的服务器。
Intel has released an AI toolkit called OpenVINO AI for the open-source audio editing program, Audacity. These plugins enable Audacity users to transcribe podcasts locally and generate music based on prompts. OpenVINO AI is only compatible with the 64-bit version of Audacity 3.4.2 for Windows and must be manually installed from the files provided on GitHub. Users on Linux or macOS now need to compile it themselves. The developers of Audacity have stated that several plugins have been provided, including three music tools: music generation, music style remixing, and music separation. The first two tools use the Riffusion model (a branch of Stable Diffusion) to generate or remix music based on text prompts. The music separation tool allows users to split existing songs into separate tracks. It can break down a song into vocals and instruments or split it into four tracks, including vocals, drums, bass, and other instruments.
Additionally, there are two tools available for speech audio content: noise suppression and transcription. These plugins allow users to respectively remove background noise and transcribe speech audio using AI. All OpenVINO AI tools run locally, eliminating the need for a connection to Intel’s servers.
微信扫一扫