Apps
Announcements
This may be going on a cool direction...I still have to do some work if I fully commit to this, but I needed to update the overall speed of this app, so... I added 5 new flavours to RE-USE, thanks to this awesome community we have on Hugginface! Props to the people who fine-tuned all of these options. (This also means that MLX support CAN be possible! I pin-pointed those models, but I don't have a Mac available, so if anyone wants to copy this one repo and do that version, feel free to do it). Added a bunch more of knobs to really tweak the output and overall speed of the recovered result (with an easy explanation to know what to tweak). Also added batch processing for people that need a "run-and-leave" solution for several broken audio. This may be a really small project for you to play with, but it is no more a "cheap clone". And maybe in the future it is much more...
Zonos3-Mini Voice Studio: Realtime voice hacking studioVoice Studio Zonos Voice Studio lets you speak text in a chosen voice and shape how that voice sounds. Speak a line with one of 30 built-in voices, or clone a voice from a short recording or upload. Shape it: gender, age, pitch, intonation, accent, texture, room, pace, loudness, and mood. Start from a preset such as BBC newsreader, gravelly old man, or late-night radio, then compare the result with the original voice. Perform a script: a different voice per line, delivery painted onto words, several takes, then export one WAV. Save the voice, cast, and script in a project you can export and import. Zonos3-mini Model Performance Very fast generation: once you clone a voice, you just enter text and it almost instantly generates the speech and starts streaming High fidelity: I've tried to clone Christopher Walken's voice with all kinds of voice cloning models in the past but this one does it the best It has two modes: Basic mode Full mode Here's what the basic voice cloning mode looks like. I don't know how they did it but this thing generates voice in realtime. The following is on my Mac M1 Max 64G: Tested on Mac and NVIDIA A4500 (Both realtime generation)
EmberWriter has come a long way since my last Pinokio post.I started building it because I wanted one place to write a novel, keep track of the world I was creating, and do something with the book when it was ready. The new feature tour follows one fictional project through that process, from a lantern going dark in the opening chapter to a submission packet you can download. Watch the updated EmberWriter feature tour (9:01, 1080p) Write the book and keep the world straight. The manuscript, character notes, and research live together. Story Atlas helps you work out whether a route or location makes sense in the next scene. Series tools help you carry established facts across books, and version history and the Recovery Center give you a way back when you need an earlier draft or a project backup. Direct the audiobook line by line. Audio Director gives you a separate performance manuscript where you can correct who’s speaking, assign character voices, and give a line its own delivery cue. You can audition a take, flag a pronunciation or performance issue, and listen again before approving it. The tour includes a multi-character excerpt rendered with EmberWriter’s Kokoro voices. Find a publisher and build the packet they ask for. The Market Directory includes 2,804 searchable book publisher records. Where we’ve researched the requirements, you can filter by genre, submission route, and packet availability. EmberWriter flags entries that still need research or require an agent, so you know where to check before spending time on a submission. In the tour, I use a publisher’s bundled settings to prepare a query and a Word attachment with the synopsis and opening chapters. Then I review the files and download the ZIP. The private authorship review lets you inspect matches against recorded AI wording, and that review stays outside the submission ZIP. Already using EmberWriter? Click Update in Pinokio, then choose DEMO to try it with fictional material. If you give it a spin, I’d like to hear which part of your writing, audiobook, or submission workflow it helped with and where you still had to reach for another tool. Install EmberWriter for Pinokio · Download all eight screenshots
A universal speech enhancement model for diverse degradations, sampling rates, and languages made by Nvidia, using Nemotron RE-USE's Hugginspace Space clone for local inference. Compatible with Windows and CUDA 12.8 GPUs
@jmw_studio_film1 check-in
PhospheneFeatured
On your Mac, make AI video with sound. Free and open source. Run LTX-2.5, Hailuo H3 and H3 Flex locally on Apple Silicon: text or image to video, lip-sync, songs, images and character training. AI agents like Claude Code and Codex can drive it too. No cloud or API key.
@bizarro
46 check-ins
Author-led fiction studio: write complete books, review every chapter, publish and narrate locally
@grumpystrongman
1 check-in
1 check-in
Higgs Audio v3 TTS + AI text director, voice cloning, podcast & audiobook (multi-speaker). 100+ languages, offline, NVIDIA GPU.
@nerual_dreming6 check-ins
Suno at home. Local AI music generation studio — full songs with vocals, lyrics, covers, and music videos. Built on ACE-Step 1.5 XL.
@nerual_dreming8 check-ins
ElevenLabs at home. Multilingual TTS with Voice Design, Voice Cloning, and end-to-end LoRA fine-tuning straight from a video or podcast. Built on VoxCPM2 by OpenBMB. 30 languages incl. Russian.
@nerual_dreming3 check-ins
AI-powered model downloader, workflow bridge, timeline video editor, and manager for ComfyUI. Features modular Video Diffusion Strategies (Wan 2.2, Wan 2.1 I2V, LTX, Hunyuan), TikTok/Reels 9:16 vertical presets, and automatic missing-node sanitization.
1-click MiniMax H3 video in ComfyUI plus H3 Studio: local T2V/I2V, references, continuation, Gallery, Director/Cinema and TR/EN. NVIDIA GPU.
@mvrdvk
24 check-ins
Build, curate, caption and clean LoRA training datasets in your browser, then train, compare and iterate — local Flask app, your files stay on your disk. https://github.com/perfectgf/lora-dataset-studio
Make images and 3D assets locally for games, 3D printing, and creative projects. Foundry gives you guided workflows, automatic setup, and export options for each use. Use it on its own or connect your coding agent with the included MCP server and companion skill. Give your agent a head start with locally generated assets, so it can spend less time and fewer tokens building from scratch. Interface in English, Simplified Chinese, and Latin American Spanish. Windows/NVIDIA preview.
@mumblebox
1 check-in
Local-first AI video editor with optional native MiniMax H3 generation on Apple Silicon.
@davidaircloud
3 check-ins
3 check-ins
Wan2GPFeatured
1-click WanGP Launcher. Super Optimized Gradio UI for AI video creation for GPU poor machines (6GB+ VRAM). Supports Wan 2.1/2.2, Qwen, Hunyuan Video, LTX Video and Flux. https://github.com/deepbeepmeep/Wan2GP
740 check-ins
YUE2 // GROOVEFeatured
YUE2 // GROOVE is the latest music studio built on the open-source Yue2 model and its inference stack. Generate high-quality full songs from style and lyrics with an editable score plan — powered by the latest YuE model — cover from audio with SheetSage2 and MERT2, refine and compare edits, and keep your works in a reusable, easy-to-manage library. You get high-quality creation with real creative control. Hardware: an NVIDIA GPU with 24 GB VRAM on Linux (YuE2's recommended setup, validated end-to-end on NVIDIA L4 hosts; a 16 GB memory budget runs everything the app can produce, and 12 GB runs the unquantized model at CFG 1.0 or for shorter songs) or an Apple Silicon Mac with 32 GB+ unified memory (where this app is developed and tested). Windows is best effort: install, launch, Cover and a full-length song verified on Windows 11 (RTX 2070, 8 GB); song generation there runs through the GGUF engine. Cards under 16 GB (and Windows) get the optional GGUF engine: the same model through yue2.cpp with an 8-bit backbone, 8.2 GB peak for a full song, measured indistinguishable from the reference rendering in a blind ABX, a different take for the same seed; the reference PyTorch configuration stays the default wherever it fits.
@bearbone
13 check-ins
Flexible Automapper for Beatsaber made for any difficulty
AI Voice Assistant — voice conversations, animated face, canvas, music generation, and more.
A local creation studio, forked from Maestro, for directing persistent worlds across video, images, sound, comics and 3D. Includes recoverable Director pipelines and optimized MiniMax H3 generation. Local AI generation needs an NVIDIA GPU (6GB+ VRAM); other computers install the editing studio with remote providers.
@theinaog
3 check-ins
