The release notes
A little better 
with every release.
Every release of Murmur, the local Mac text-to-speech app: new document workflows, voice design, projects, and reliability fixes. All on-device.
Free plan with Kokoro voices, scripts up to 1,500 characters · 3-day Pro trial, then $49 once
Release history
18 releases · Newest first
Longer voice samples and restored voice profiles
FixVoice Cloning
- Voice samples longer than a model's recommended range no longer fail as if the recommendation were a hard limit.
- My Voices now shows each sample's measured duration and flags model-specific guidance or incompatibility before generation.
- Murmur restores missing duration metadata for legacy voice profiles and accepts imports up to 3 minutes 30 seconds.
- Local speech bridges now retain the previous session's log instead of erasing it on startup.
Model setup without administrator access
FixModel Setup
- Fixed downloadable model installation on Macs without Homebrew or Xcode Command Line Tools.
- Murmur now securely installs private Python runtimes inside Application Support without administrator access.
- Applied the installer repair to Qwen3-TTS, Voice Design, Chatterbox, OmniVoice, and Fish Audio S2 Pro.
Complete narration with Fish Audio S2 Pro
FixFish Audio
- Fixed Fish Audio S2 Pro generations stopping before the complete text was spoken.
- Increased Fish S2 Pro's generation allowance and split longer sentences at safe word boundaries.
- Added explicit detection for generation-limit exhaustion so incomplete audio is never silently returned.
More reliable voice cloning and recovery
ImprovedVoice Cloning
- Improved voice cloning reliability for Fish Audio S2 Pro, SparkTTS, Qwen3-TTS, OmniVoice, and Chatterbox.
- Added clear validation for reference duration, exact transcripts, unreadable files, and available unified memory.
- Fixed reference-audio imports from the file picker, including MP3 and WAV conversion.
- Improved cancellation, recovery, long-text generation, and concurrent request handling.
- Replaced generic Foundation failures with actionable local bridge diagnostics.
- Removed the bundled model-download credential from public app packages.
CLI, MCP, and a more reliable production workflow
NewAutomationProjects
- Added local CLI and MCP automation so Terminal tools and coding agents can generate speech through Murmur.
- Added job progress, cancellation, batch generation, model and voice discovery, and safer local automation controls.
- Improved Projects, timeline playback and export, persistent queues, history, voice cloning, and model management reliability.
- Refined the app interface and added complete automation documentation on the Murmur website.
Qwen3 runtime reliability
FixQwen3Voice Design
- Fixed the local runtime path used by Qwen3-TTS Base and Voice Design.
- Improved setup diagnostics for MLX-backed voice generation.
Model detection and history reliability
FixModelsHistory
- Fixed installed models occasionally appearing missing after relaunching the app.
- Improved generated-audio history retention and storage behavior.
EPUB import for local audiobook workflows
NewImportDocuments
- Added EPUB document import for DRM-free books.
- Murmur extracts readable chapters locally so they can be turned into audio without uploading the book text.
Favorite community voices
NewVoicesExplore
- Added favorites for community voices in Explore.
- Star voices, filter to favorites, and quickly return to voices you plan to use again.
Projects workspace
NewProjectsWorkspace
- Added the Projects workspace for longer scripts and multi-speaker production.
- Added speakers, timeline editing, export flow, Fish prompt chunking, and the MLX speech bridge for local project generation.
Voice Design workspace
NewVoice Design
- Added the Voice Design workspace with guided prompt presets.
- Added OmniVoice setup, language selection, generated voice previews, and saving designed voices into My Voices.
Python backend installation fix
FixBackend
- Fixed installation failures on Macs where Python 3.13 or newer is the default.
- Murmur now automatically picks up or installs Python 3.11 for the Fish Audio and Qwen3-TTS backends.
Voice cloning reliability
ImprovedVoice Cloning
- Improved Spark voice cloning reliability.
- Improved update diagnostics for easier troubleshooting.
Maintenance update
Fix
- Bug fixes and improvements.
Checkout redirect fix
FixCheckout
- Fixed website checkout redirects after purchase.
- Updated the in-app buy link to the live murmurtts.com domain.
Maintenance update
Fix
- Bug fixes and improvements.
Update validation fix
FixUpdates
- Fixed an update validation issue for the 1.0.1 release.
- Served the signed archive from a fresh immutable URL.
Murmur 1.0 — turn any text into natural speech
LaunchVoicesPrivate
- The first public release of Murmur, a text-to-speech app made for the Mac.
- Paste or type any text, pick a voice, and listen to a natural-sounding read instantly.
- Everything happens right on your Mac — your text never leaves your computer, and it works completely offline.
- Choose from a library of expressive voices to find the tone that fits what you're listening to.
- Speed up or slow down the reading to match the pace you prefer.
- Built-in player to play, pause, and scrub through your audio without leaving the app.
- Every read is saved to your history, so you can revisit and replay anything you've made.
- Export your audio to keep it, share it, or drop it into another project.