Skip to content

Speak naturally.Keep it private.

Turn your voice into polished text in any app — processed locally on your PC or Mac through key guarantees like No uploaded audio, 0 cloud transcripts, and Zero data selling

Murmur ready · Click to dictate
Network simulation:
Target app: Slack
What you said, unedited

hey team quick update on the latency benchmarks real time factor dropped to zero point two please test and report any bugs in eng channel

Hey team, quick update on the latency benchmarks: • Real-time factor dropped to 0.20x with sub-180ms latency • DirectML and Metal GPU offloading are active • Please test your dictations and report any edge cases in #eng-bugs
whisper.cpp DirectML / Metal
Latency: ~172msNetwork egress: 0 bytes
Universal Application Support

Works instantly anywhere you can place a cursor

No browser plugins, no custom integrations, and zero clipboard overwrites. Murmur simulates native OS keyboard events directly into the frontmost window.

{ }
CursorAI Code Editor
//
VS CodeCode Editor
#
SlackTeam Chat
LinearIssue Tracking
GitHub DesktopVersion Control
>_
TerminalmacOS Shell
RaycastProductivity
$
WarpModern Terminal
XcodeApple Development
~
iTerm2Terminal
ij
IntelliJ IDEAJetBrains IDE
PostmanAPI Testing
Docker DesktopContainers & DevOps
py
PyCharmPython IDE
=
Sublime TextText Editor
v
NeovimModal Editor
{ }
CursorAI Code Editor
//
VS CodeCode Editor
#
SlackTeam Chat
LinearIssue Tracking
GitHub DesktopVersion Control
>_
TerminalmacOS Shell
RaycastProductivity
$
WarpModern Terminal
XcodeApple Development
~
iTerm2Terminal
ij
IntelliJ IDEAJetBrains IDE
PostmanAPI Testing
Docker DesktopContainers & DevOps
py
PyCharmPython IDE
=
Sublime TextText Editor
v
NeovimModal Editor
N
NotionDocs & Knowledge
ClaudeAI Intelligence
ChatGPTAI Assistant
FigmaInterface Design
Apple MailEmail Client
ObsidianMarkdown Notes
Google DocsCollaborative Docs
ArcModern Browser
Apple NotesQuick Notes
::
DiscordCommunity Chat
Google SheetsSpreadsheets
ChromeWeb Browser
X
ExcelData & Finance
🧭
SafariApple Browser
TelegramEncrypted Chat
AirtableConnected DB
N
NotionDocs & Knowledge
ClaudeAI Intelligence
ChatGPTAI Assistant
FigmaInterface Design
Apple MailEmail Client
ObsidianMarkdown Notes
Google DocsCollaborative Docs
ArcModern Browser
Apple NotesQuick Notes
::
DiscordCommunity Chat
Google SheetsSpreadsheets
ChromeWeb Browser
X
ExcelData & Finance
🧭
SafariApple Browser
TelegramEncrypted Chat
AirtableConnected DB
Zero browser plugins or extensions
Preserves your existing clipboard history
Sub-180ms OS synthetic key injection
The Core Philosophy

Yourvoiceisyourfastesttool.Turnunfilteredthoughtsintopristinewriting,code,anddecisionswithouttouchingthecloud.

Whisper AI inference runs directly on your Apple Silicon Neural Engine or Windows DirectML GPU with zero cloud subscriptions, zero telemetry, and complete offline privacy.

whisper.cpp engine

100% On-Device Compute

All acoustic feature extraction and token decoding run directly in local GPU/Neural Engine memory. No cloud roundtrips, no API keys, and no rate limits.

Air-gap verified

Zero Telemetry or Audio Logs

Your microphone audio never touches a remote server. Verified by packet inspection tools like Wireshark and LuLu. 0 bytes outbound network egress.

Sub-180ms latency

Universal OS Injection

Dictate into Cursor, Slack, Notion, Xcode, or Terminal at the OS level. Murmur simulates native key events directly without overwriting your clipboard history.

Interactive Voice Lab

Test Murmur in your browser.

Experience how Murmur turns unstructured messy speech into formatted writing while stripping filler words with zero cloud latency.

Select a scenarioInstant decode
Local system telemetry
Processing latency184 ms
Cloud egress0 bytes
VRAM footprint~380 MB
Software costFree (MIT)

Backend Refactor Note

Coding & Architecture
GPU-accelerated voice detection
What you said:

we need to refactor the payment gateway to use exponential backoff with jitter and add an idempotent key to prevent double charging on stripe webhooks

// Architecture Decision: Payment Gateway Resiliency 1. Implement exponential backoff with Full Jitter algorithm 2. Enforce Idempotency-Key headers on all Stripe webhook handlers 3. Guarantee zero duplicate billing during network timeouts
Pasted directly into active cursor position
Dated, Reproducible Benchmarks

Beat the cloud on latency.

Local AI is only compelling if it feels instantaneous. We engineered Murmur in native Rust to outpace cloud WebSocket pipelines at every stage.

Perceived Speed

Tail Latency (Speech End → Injected)

3.2x Faster
Murmur (Native On-Device)140 – 190 ms
Zero network hop · Direct GPU decode
Cloud Dictation (Wispr / Cloud APIs)450 – 850 ms
WebSocket upload + Cloud queue + TLS overhead

Cloud dictation pays a mandatory network round-trip penalty (DNS + TLS handshake + WebSocket upload + cloud inference queue). Murmur decodes inside GPU VRAM immediately.

Test Configuration: M3 Max / Windows RTX 4070 · Whisper Base Q5_0 · 15s utterancesReproducible via cargo bench
Adaptive Writing Styles

Make Murmur sound like you.

Murmur adapts to how you write in different apps. Set a different style for messages, work chats, emails, and code editors—computed 100% locally on your machine.

Select Sample Voice:
Raw Spoken Voice Input

hey are you free for lunch tomorrow let's do 12 if that works for you

Formatted as Natural Casual100% On-Device Rewrite

Hey, are you free for lunch tomorrow? Let’s do 12 if that works for you.

Automatically applied when target application focused0ms Network round-trip
Smart Context Engine

One voice input. Formatted for every app.

Murmur senses the frontmost active window and shapes tone, punctuation, and layout to fit the medium.

Raw Spoken Speech (Single Take)

hey mark we just merged the auth token rotation fix to main let me know if staging looks good thanks

Whisper detected: EN
Formatted Output for Slack & DiscordCasual structure with bullet points and mentions
Hey Mark. Just merged the auth token rotation fix to main. • Token invalidation grace period is set to 15 mins • Unit & integration tests are all green Could you give staging a quick sanity check? Thanks!
Power-User Customization

Own your customization.

Generic cloud dictation gives you one rigid output style. Murmur gives you complete local automation, vocal snippets, and portable dictionaries.

Live Configuration Preview

Voice-Triggered Text Snippets & Macros

100% Local Logic

Define vocal expanders. When you say 'bug template', 'schedule link', or 'invoice address', Murmur replaces the phrase with your structured schema immediately.

Vocal Trigger / Voice Input

“insert bug template”

Target Window Output
### [Bug]: Title
- **Steps to Reproduce:**
  1. Open dashboard settings
  2. Toggle Air-Gap mode
- **Expected:** Sockets close immediately
- **Actual:** Verified 0 bytes egress
- **Environment:** Murmur v0.1.0 · macOS Sonoma
Built for High-Trust Workflows

Features that solve real daily jobs.

Built for developers, lawyers, healthcare professionals, and founders who cannot—and will not—stream sensitive voice data to the cloud.

Clean-up and formatting

“Speak the messy first draft. Paste the polished version.”

Speak freely with false starts and filler words. Murmur strips ums, ahs, and repetitions, injects proper punctuation, and outputs clean paragraphs ready to send.

For:Busy operators, executives & consultants
Smart cleanupAuto-punctuation
Global hotkey + universal insertion

“One shortcut works wherever you write.”

Press Option+Space (macOS) or Alt+Space (Windows) in any text field. Speak naturally and text pastes directly at your cursor via native OS input injection.

For:Slack, Cursor, Word, Notion & Gmail users
Universal hook⌥Space / Alt+Space
Fully local processing

“Your client notes and private thoughts stay yours.”

Inference runs 100% on your local CPU/GPU using whisper.cpp. Audio is held only in RAM and deleted on decode. Zero cloud telemetry, zero remote databases.

For:Lawyers, therapists, founders & developers
Zero cloud telemetry0 bytes sent
Offline dictation

“Dictate on a plane, in a hotel, or anywhere Wi-Fi fails.”

Whisper weights execute completely offline on your device. Transcribe on 35,000-ft flights, in subways, or in secure air-gapped rooms with zero internet connection.

For:Frequent travelers & air-gapped environments
Offline mode100% local model
Custom vocabulary / dictionary

“Teach it client names, code terms, medications, or product names.”

Add proprietary names, code symbols, technical acronyms, or medication terms. Murmur biases recognition weights to achieve flawless accuracy.

For:Engineers, medical teams & specialized domains
Custom dictionaryPhonetic biasing
Local language models

“Private dictation in the languages you actually use.”

Automatic language detection across 99 languages. Dictate in Spanish, German, Japanese, French, or English without cloud transmission or reconfiguring settings.

For:Multilingual teams & global professionals
Multilingual99 languages
Three-Step Workflow

How Murmur works in three steps.

No window switching or manual copying. Speak and your thoughts become structured writing.

01

Press your global shortcut

Tap Option+Space on macOS or Alt+Space on Windows from any focused application to trigger the minimal pill.

Instant FocusStep 01
02

Speak at your natural pace

Speak naturally in your preferred tone. Local Whisper models decode audio streaming chunks concurrently on your GPU.

GPU StreamingStep 02
03

Receive formatted text instantly

Release the shortcut. Murmur clears filler sounds, structures sentences, and pastes text directly into your document.

Instant PasteStep 03
Architecture vs. Policy

Cloud dictation vs. Local dictation.

Cloud tools protect data with privacy policies. Murmur protects data by keeping your dictation on your device in the first place.

The Local-First Difference: You don't have to trade speed, usability, or smart formatting to get complete air-gapped privacy.

CapabilityMurmurLocal-First (Free MIT)Wispr FlowSuperwhisper
Architecture & Data Boundary100% On-Device (Zero cloud upload)Cloud transcription (Audio uploaded)Hybrid (Cloud for advanced)
Privacy Guarantee MechanismPhysical architecture (Zero bytes sent)Policy & settings basedPolicy based
Offline Dictation (Planes & Travel)
Universal Global Hotkey (Any App)
Filler Word Stripping & PunctuationInstant local formattingCloud LLMLocal / Cloud
Custom Jargon & Phonetic BiasingUnlimited on-device dictionaryCloud managedPro tier only
macOS & Windows ParityNative macOS & Windows (Metal / DirectML)macOS & Windows previewmacOS only
LatencyUnder 180 ms (Local GPU)300 to 700 ms (Network ping)250 ms
Pricing & LicenseFree & Open Source (MIT)$144 / year subscription$200 lifetime
Time & Value ROI Calculator

Calculate your speaking speed dividend.

Human speech flows at ~160 words per minute while average typing tops out at ~40 words per minute.

3.5 hrs / day
0.5 hr (Light)4.0 hrs (Average)8.0 hrs (Heavy)
$65 / hr
$20 / hr$100 / hr$250 / hr
Monthly Hours Saved
57 hrs~13.1 hours reclaimed weekly
Work Days Gained / Year
+82 daysAnnual productivity dividend
Annual Value Created
$42,656$0.00 Murmur software cost (Free)
Local Whisper Architecture

Select the right model for your device.

Murmur runs 100% on device with Whisper GGML models. Whisper Small (190 MB) is included free by default; download any model weights with one click.

Whisper Small (q5_1)

Default Recommended

The official default for all users. Optimal balance of sub-200ms speed, low memory footprint, and accurate punctuation.

Optimal Hardware & Use-Case:Software developers, daily dictation, technical notes, bilingual typing
Latency160 ms
Disk Size190 MB
RAM Usage450 MB
Accuracy98.5%
Verified Community Feedback

Trusted by engineers, clinicians, and writers.

Practical feedback from professionals who rely on on-device speech recognition daily.

Engineering

Murmur completely replaced cloud dictation tools for my daily engineering work. I write code comments, PR reviews, and Slack messages with my voice. Being 100% offline and free is an incredible achievement.

E

Elena Rostova

Staff Software Engineer

Healthcare

As a physician, HIPAA compliance is strictly non-negotiable. I cannot use cloud speech tools. Murmur runs locally on my laptop with zero network traffic and recognizes medical terminology reliably.

J

Dr. Julian Vance

Clinical Neurologist

Writing

I dictated over 40,000 words of my latest manuscript with Murmur. The filler word removal cleans false starts without interrupting my natural drafting rhythm.

M

Marcus Sterling

Author & Essayist

Architecture

Alt+Space on Windows with DirectML is instantaneous. The real time factor is under 0.20x. My wrist typing strain has completely disappeared.

S

Siddharth Patel

Lead Architect

Product

Murmur is faster than paid cloud alternatives because it eliminates the network latency roundtrip entirely. Solid native engineering in Rust.

C

Chloe Dubois

Product Manager

DevOps

The custom dictionary feature is essential for our proprietary APIs, microservice acronyms, and team names. Whisper recognizes them accurately every time.

D

Devon Miller

Infrastructure Lead

Legal

Attorney-client privilege cannot survive unvetted cloud audio streaming. Murmur's local air-gapped processing is the only dictation architecture our compliance committee cleared for confidential litigation briefs.

L

Sarah Lin, Esq.

Partner, Tech IP Litigation

AI Research

Conducting frontier AI research means our unreleased papers and hypotheses are sensitive IP. Running whisper.cpp on an Apple M3 Max with zero telemetry allows me to dictate notes freely.

A

Mateo Alvarez

ML Research Scientist

Mental Health

Clinical therapy notes require absolute patient confidentiality. Knowing voice audio stays exclusively in RAM and vanishes upon paste gives me and my patients total peace of mind.

H

Dr. Hannah Weiss

Clinical Psychologist

Executive

Our security policy banned cloud transcription bots company-wide. Murmur passed our internal packet inspection and security audit with flying colors. 100% on-device is the future.

P

Priya Nair

VP of Engineering

Engineering

Writing git commit messages, GitHub PR reviews, and technical documentation hands-free without any cloud lag has doubled my daily throughput. The shortcut is second nature now.

O

Liam O'Connor

Senior Full-Stack Engineer

Finance

Sub-180ms latency on Windows with an RTX 4080. It's the only dictation software that keeps pace with rapid market analysis and real-time trading journal entries without stutters.

X

Alexander Chen

Quantitative Trader

Ready to Dictate

Download Murmur. Free forever.

No credit card, account, or cloud telemetry required. Installs and runs in under 60 seconds.

macOS

Detected Device
macOS 13.0 or later (Apple Silicon & Intel)

Native Metal GPU acceleration on Apple Silicon (M1, M2, M3, M4) and Intel Macs. Runs quietly in your menu bar.

Windows

Windows 10 & 11 (64-bit)

Native DirectML & CUDA acceleration for NVIDIA, AMD, and Intel GPUs. Operates seamlessly in your system tray.

Install via Command Line Package Managers

Windows Winget
winget install WebProdigies.Murmur
Windows PowerShell
irm https://murmur.app/downloads/murmur-setup.exe -OutFile murmur-setup.exe; .\murmur-setup.exe
macOS Homebrew
brew install --cask murmur
Source / GitHub Clone
git clone https://github.com/webprodigies/murmur.git
Disk Space190 MB for default Whisper Small model
GPU AccelerationApple Silicon Metal & Windows DirectML
Privacy VerifiedZero network egress · 100% On-Device
Questions & Answers

Frequently asked questions.

Everything you need to know about Murmur performance, privacy, and local architecture.

Never. All audio capture, Voice Activity Detection (VAD), Whisper model inference, and text formatting run exclusively on your local CPU and GPU. The only network calls Murmur makes are the initial model weight download from official Hugging Face mirrors and an optional update check.