Moonshine

Fast and accurate automatic speech recognition model for edge devices.

CommonProductProductivityAutomatic Speech RecognitionEdge Computing

Moonshine is a suite of speech-to-text models optimized for resource-constrained devices, making it ideal for real-time, on-device applications such as live transcription and voice command recognition. It outperforms the OpenAI Whisper model of the same size in word error rate (WER) on test datasets used in the OpenASR leaderboard maintained by HuggingFace. Additionally, Moonshine's computational requirements vary with the length of the input audio, allowing for quicker processing of shorter audio compared to the Whisper model, which processes everything in 30-second chunks. Moonshine processes 10-second audio segments at a speed five times faster than Whisper while maintaining the same or better WER.

Real-time transcription: Suitable for live transcription and voice command recognition.
Optimized word error rate: Outperforms the Whisper model across multiple datasets.
Fast processing: Processes shorter input audio five times faster than Whisper.
Multi-platform support: Compatible with Torch
TensorFlow
and JAX backends.
Flexible deployment: Operates on resource-constrained edge devices.
Easy installation: Provides detailed installation guides and virtual environment setup.
Model selection: Offers two model choices: 'moonshine/tiny' and 'moonshine/base'.

\Moonshine is designed for users who need fast and accurate voice recognition on resource-constrained devices
such as developers
businesses
and individuals requiring real-time voice transcription services. It is particularly suited for scenarios where voice interaction is needed on mobile or IoT devices.\

Developers can leverage Moonshine to add real-time voice recognition features to mobile applications.
Companies can integrate Moonshine into customer service systems to provide voice-to-text services.
Individuals can use Moonshine to transcribe audio records of meetings or lectures.

1. Install uv for Python environment management.
2. Create and activate a virtual environment: use 'uv venv env_moonshine' and 'source env_moonshine/bin/activate'.
3. Install the Moonshine package
choosing an appropriate backend (Torch
TensorFlow

Visit

Moonshine Visit Over Time

Monthly Visits

515580771

Bounce Rate

37.20%

Page per Visit

5.8

Visit Duration

00:06:42

Moonshine Visit Trend

Moonshine Visit Geography

Moonshine Traffic Sources

Moonshine Alternatives

Workers AI — Run machine learning models on Cloudflare's global network.

InternationalSelection•Machine Learning•Edge Computing

whisper-ner-v1 — An advanced model for joint speech transcription and entity recognition.

Productivity•Speech Recognition•Entity Recognition

WhisperNER — Unified open-source named entity and speech recognition model

Programming•Automatic Speech Recognition•Named Entity Recognition

Quantized Llama — An efficient, lightweight Quantized Llama model that enhances performance on mobile devices while reducing memory usage.

Productivity•Quantization•Mobile Devices

TEN-framework — The next-generation AI agent framework, the world's first true real-time multimodal AI agent framework.

Programming•Cross-platform•Plugin System

120

Ministral-8B-Instruct-2410 — A high-performance language model that supports local intelligence and on-device computation.

Programming•Large Language Model•Local Intelligence

Rev AI — The world's most accurate AI voice transcription service

InternationalSelection•Voice Recognition•Transcription Service

144

Whisper large-v3-turbo — Efficient automatic speech recognition model

Productivity•Automatic speech recognition•Speech translation

516

Llama 3.2 — Open-source AI model that can be fine-tuned, distilled, and deployed.

Productivity•Machine Learning•Open Source

174

CrisperWhisper — Word-level automatic speech recognition model

Productivity•Automatic Speech Recognition•Verbatim Transcription

JotMe — Real-time translation and meeting recording by AI assistant on Google Meet

Productivity•AI Translation•Meeting Recording

MBox AI Meet — A Google Meet extension for real-time transcription and meeting summaries.

Productivity•Meeting Management•Real-time Transcription

162

Friend — Smart wearable necklace, revolutionizing the way conversations are recorded and managed.

Others•Wearable Device•Open-Source

222

iPadOS 18 — iPadOS 18 brings a whole new way of working and personalized experience.

Productivity•Smart Ink•Personalization

294

seed-tts-eval — A testing dataset for evaluating a model's zero-shot speech generation capability

OpenSource•Speech Synthesis•Automatic Speech Recognition

1176

Grounding DINO 1.5 API — Advanced open-world object detection model series

Image•Object Detection•Open World

534

VILA — A multi-image visual language model with training, inference, and evaluation solutions, deployable from cloud to edge devices (such as Jetson Orin and laptops).

Image•Visual Language Model•Video Understanding

792

Octopus-V2 — Octopus-V2-2B is a 2B LLM running on mobile devices with performance exceeding GPT-4.

Productivity•Large Language Model•Mobile Device Optimized

3120

Best AI Websites & Tools