NexSub
Offline real time subtitle translation using local Whisper models for any video source
About NexSub
NexSub is a desktop application that captures system audio and generates real time subtitles in over 100 languages, all without an internet connection. It works by running speech recognition locally via Whisper models on your GPU, then translating and displaying subtitles with roughly 1.5 seconds of latency. You can use it with YouTube, Netflix, Zoom, local video files, or anything else that produces audio on your computer.
The offline angle is the main selling point. Most subtitle and translation tools send audio to cloud servers, which means latency, usage limits, and the question of who else is listening. NexSub keeps everything on your machine. Once you download the models, you can unplug from the network entirely and the tool keeps working. That makes it attractive for corporate environments handling confidential content, travelers without reliable connectivity, and anyone privacy conscious about what gets uploaded.
The way it works is straightforward. NexSub hooks into your system's audio output, runs it through Whisper for transcription, then passes the text to a local translation model. Subtitles appear in a draggable overlay you can position anywhere on screen. You can display bilingual subtitles showing both the original language and the translation, which is useful for language learners who want to follow along in both.
Export is baked in. You can save subtitles to SRT or VTT files for archiving or embedding into video later. This turns NexSub into a transcription tool as well, capturing meeting notes or lecture content that you can edit afterward.
Hardware requirements are the main constraint. You need an NVIDIA GPU with at least 6 GB of VRAM, realistically a GTX 1660 or RTX 2060 at minimum. Recommended is an RTX 3060 or better with 8 GB or more. If you're on integrated graphics or an older card, the models won't run fast enough for real time use.
Pricing is a one time $9 purchase for lifetime access. There are no subscriptions, no per minute fees, and no server costs because nothing runs in the cloud. For users who process a lot of video or run it daily, the flat fee pays for itself quickly compared to subscription translation services.
NexSub fits language learners who want subtitles while watching foreign content, remote teams communicating across languages, researchers transcribing interviews, and anyone who needs reliable subtitles in places where internet is spotty or where privacy matters more than convenience.
Key Features
- 100 percent offline operation with local Whisper models
- Real time bilingual subtitle display
- System audio capture for any video source
- Export to SRT and VTT formats
- Support for over 100 languages
- Draggable subtitle overlay positioning
Pros & Cons
What we like
- Completely offline after initial model download
- One time $9 payment with no subscriptions
- Works with any audio source, not just specific platforms
- Bilingual display helps language learners follow along
Room for improvement
- Requires NVIDIA GPU with at least 6 GB VRAM
- No free tier or trial to test before buying
- Latency of around 1.5 seconds may lag behind fast dialogue
- Windows only based on current documentation
Frequently Asked Questions
What is NexSub?
Does NexSub require an internet connection?
What are the hardware requirements?
How much does NexSub cost?
Best For
Featured in
Alternatives to NexSub
View allCartesia
Ultra-low-latency real-time text-to-speech powered by the Sonic model, built for live voice AI agents
LOVO AI
AI voice generator and video studio with 500+ voices across 100+ languages, plus voice cloning
ElevenLabs
The voice cloning and text-to-speech service everyone benchmarks against
WellSaid Labs
Enterprise text-to-speech with studio-quality AI voice avatars trained on consenting voice actors
Reviews (10)
Quietly excellent
NexSub solves a real problem for me without making a fuss about it. What stands out is how it handles support for over 100 languages.
Recommended without reservation
Came to NexSub after getting frustrated with what I had before. Got real value out of export to srt and vtt formats. What stands out is how little babysitting it needs. Recommending it to people in a similar spot.
Powerful once it clicks
Three months of NexSub later, here is what holds up. What stands out is how it handles one time 9 payment with no subscriptions. My only gripe is latency of around 1.5 seconds may lag behind fast dialogue. Recommending it to people in a similar spot.
Good, with a few caveats
Found NexSub on a Show HN thread and I am glad I clicked. The system audio capture for any video source is more useful than I expected. The catch is no free tier or trial to test before buying.
Solid but not perfect
Have been running NexSub for a while, here is where I land. Got real value out of bilingual display helps language learners follow along. The catch is latency of around 1.5 seconds may lag behind fast dialogue. It earns its place in my stack.
Solid daily driver
Three months of NexSub later, here is what holds up. Where it really wins is completely offline after initial model download. It does what it says, which is rarer than it should be. Worth it for what I get out of it.
Decent with some rough edges
Three months of NexSub later, here is what holds up. What stands out is how it handles completely offline after initial model download. Support actually answered when I had a question, which surprised me. It would be a five if not for latency of around 1.5 seconds may lag behind fast dialogue. Glad I made the switch.
Solid daily driver
Three months of NexSub later, here is what holds up. Got real value out of support for over 100 languages. Mostly using it for watching foreign language shows with real time translation. Easy yes for anyone weighing the same trade offs.
Worth a look
Came to NexSub after getting frustrated with what I had before. What stands out is how it handles draggable subtitle overlay positioning. Found it works best for generating srt files from local video files offline.
Good, with a few caveats
Three months of NexSub later, here is what holds up. Performance has been steady even when I lean on it hard. It slotted into my routine without much fuss. Mostly using it for learning a language by following bilingual subtitles. The catch is requires nvidia gpu with at least 6 gb vram. Easy yes for anyone weighing the same trade offs.
Related Tools
WellSaid Labs
Enterprise text-to-speech with studio-quality AI voice avatars trained on consenting voice actors
Cartesia
Ultra-low-latency real-time text-to-speech powered by the Sonic model, built for live voice AI agents

Contextli
Context-aware AI voice assistant for Email, Slack, and every app
LOVO AI
AI voice generator and video studio with 500+ voices across 100+ languages, plus voice cloning