Tag
AMBIT是一个统一框架，通过将自回归场景规划与rectified-flow渲染耦合，实现第一阶Ambisonics（FOA）空间音频的生成与编辑，并利用FOA感知的连续VAE与on-policy自蒸馏来保持空间结构和声学保真度。
Audionaut is a free, open-source multitrack audio editor built in C++ on JUCE, running natively on Windows, macOS, and Linux, with MCP support that lets Claude and other AI agents drive edits, stem separation, and audio analysis.
Robot Voice Bridge is a native Mac app that integrates ElevenLabs for streamlined voiceover sessions, enabling users to generate, edit, and drop audio directly into editing software.
Ardour 9.8 introduces key and scale visualization for MIDI tracks, improves session recovery and clip recording, and adds an arpeggiation option in MIDI tools.
Tencent has open-sourced AuK, a 1.5B foundation model for speech generation and editing that supports zero-shot TTS, content editing, and more through a unified natural-language instruction interface.
Vocal Slice is a desktop application for audio editing that uses on-device Whisper transcription to allow users to select and export clips by highlighting text in the transcript, designed for podcasters and voice professionals.
iVox is the first app dedicated to recreating 1980s tape-edit effects for audio editing.
Redactify is a tool that automatically censors profanity in video and audio content.
MMAE is a comprehensive benchmark for multitask audio editing that evaluates AI's ability to precisely modify existing audio clips via natural language instructions, with current models achieving under 5% exact match rate.
MMAE is a comprehensive benchmark for instruction-based audio editing across multiple modalities and complexity levels, revealing significant gaps in current model capabilities.
Adam Solove describes Ducking, a custom podcast editor he built featuring a magnetic timeline, multiplayer editing, and collaboration tools purpose-built for spoken-word audio workflows. The post explores UI design decisions that improve on traditional DAW interaction models for podcast editing.