WhisperFusion

by Collabora

Ultra-low-latency speech→LLM pipeline: WhisperLive + Mistral + TensorRT.

TL;DR

Ultra-low-latency speech→LLM pipeline: WhisperLive + Mistral + TensorRT.

Best for voice-agent prototypes that need 500ms end-to-end speech→reply on a single GPU. Pricing: free.

Category
Open source
License
—
Stars
—
Last push
—
Pricing
free
Platforms
Linux, Docker

What it is

Combines WhisperLive (ASR), Mistral-7B (LLM), and SileroVAD into one Triton-served pipeline. Demonstrates sub-second turn-taking on an RTX 4090. MIT.

Best for: Voice-agent prototypes that need 500ms end-to-end speech→reply on a single GPU.
Watch out for: Requires NVIDIA GPU + TensorRT-LLM; English-tuned; heavy Docker image.

Install / use

git clone https://github.com/collabora/WhisperFusion

Features

Speaker diarizationNo
Word-level timestampsYes
Streaming / real-timeYes
Languages supported99
HIPAA eligibleNo

WhisperFusion vs Whipscribe

FeatureWhisperFusionWhipscribe
CategoryOpen sourceTranscription APIs
Pricingfreefree beta
Speaker diarizationNoYes
Word timestampsYesYes
StreamingYesNo
Languages9999
PlatformsLinux, DockerWeb, API, MCP

Alternatives to WhisperFusion

OpenAI Whisper
OpenAI

The reference open-source multilingual ASR model from OpenAI.

OSS · MIT ★ 98.1k
whisper.cpp
Georgi Gerganov

C/C++ port of Whisper — runs on anything, from a Raspberry Pi to Apple Silicon.

OSS · MIT ★ 48.8k
faster-whisper
SYSTRAN

4× faster than reference Whisper using CTranslate2 — production sweet spot.

OSS · MIT ★ 22.3k

Whipscribe is a managed faster-whisper + whisperX service. If you want transcripts without running infrastructure, paste a URL or drop a file in the form below — you'll have a transcript in seconds.