Coming Soon

The VoxBar Catalog

Choose from 16 offline, privacy-first AI transcription models optimized for Windows. From zero-VRAM CPU models to high-accuracy multi-billion parameter GPU behemoths.

VoxBar™ Pro

Voxtral Real-Time 4B

VRAM: ~8-10GB Offline

The flagship offline real-time model.

VoxBar™ Pro Native

Voxtral Real-Time 4B (native)

VRAM: ~9.6GB Offline

Native HuggingFace pipeline implementation.

VoxBar™ AI

Canary Qwen 2.5B

VRAM: ~6-8GB Offline

State-of-the-art SALM architecture.

VoxBar™ Ultra

Parakeet TDT 0.6B v2

VRAM: ~2GB Offline

High-speed Parakeet model for exceptional accuracy.

VoxBar™ Flash

Canary 1B Flash

VRAM: ~3-4GB Offline

Fast, lightweight Canary variant.

VoxBar™ Moonshine

Moonshine v2

VRAM: 0GB (CPU) Offline

CPU-only model requiring zero VRAM.

VoxBar™ Whisper

Faster-Whisper (CTranslate2)

VRAM: 0-2GB Offline

The classic Whisper engine optimized with CTranslate2.

VoxBar™ Whisper+

Distil-Whisper

VRAM: 0-2GB Offline

Distilled Whisper for faster execution.

VoxBar™ Whisper Pro

Faster-Whisper Large

VRAM: 0-2GB Offline

Maximum accuracy Whisper variant.

VoxBar™ Kyutai

Kyutai Moshi

VRAM: ~2.5GB Offline

Real-time subtitling (0.5s delay).

VoxBar™ Nemotron

Nemotron Speech ASR 0.6B

VRAM: ~2GB Offline

NVIDIA Nemotron Speech model.

VoxBar™ Kyutai 2.6B

VoxBar Kyutai 2.6B (6GB VRAM)

VRAM: ~5.4GB Offline

Meeting and recording transcription (~4s delay).

VoxBar™ GLM

GLM-ASR-Nano 1.5B

VRAM: ~8GB Offline

Zhipu AI GLM model.

VoxBar™ Qwen ASR

Qwen3-ASR 1.7B

VRAM: ~6-8GB Offline

Alibaba Qwen3 1.7B ASR.

VoxBar™ Qwen ASR Mini

Qwen3-ASR 0.6B

VRAM: ~2-3GB Offline

Lightweight Qwen3 0.6B variant.

VoxBar™ SenseVoice

SenseVoice-Small

VRAM: ~1-2GB Offline

FunAudioLLM SenseVoice model.