Skip to content
NanoVoice

NanoVoice changelog

New features, improvements, and fixes for NanoVoice on Mac.

Latest release

v1.3.0

View on GitHub

What's New in 1.3.0

  • Improve on-device Apple Speech transcription on macOS 26 and 27, with language model readiness, download progress, and cancellation in Settings.
  • Reduce memory usage during local formatting and improve model loading, cancellation, and recovery.
  • Improve Apple template formatting and explain when a transcript exceeds the model’s context limit.
  • Fix Recordings window dragging, appearance, and keyboard navigation, including immediate Up/Down navigation when opening the window or searching.

What's New in 1.2.9

  • Reduce idle memory usage by releasing local speech and formatting models after a minute without use.
  • Move and resize the Recordings window, with size and position remembered.
  • Add a Parakeet recommendation in transcription settings.

v1.2.8

View on GitHub

What's New in 1.2.8

  • Search recordings by title, transcript, generated output, template, language, or recording type; pin and rename important recordings.
  • Correct a transcript and re-run it with any template without recording again; cancellation and backend errors keep the previous output safe.
  • See actual installed model sizes, active and recommended models, proactive setup guidance, and consistent deletion controls for local models.
  • Stop offering to record Discord as a meeting.

v1.2.7

View on GitHub

What's New in 1.2.7

  • View your license activation count and reset old installations from License settings.
  • Improved license recovery and clearer activation error messages.

What's New in 1.2.6

  • Fixed Whisper models being reported unavailable when an installed model uses a legacy or abbreviated name
  • Added approximate download size and active-memory requirements for Whisper, Parakeet, and built-in local LLM models
  • Fixed global shortcut handling and labels on non-US keyboard layouts

What's New in 1.2.5

  • Added optional automatic Markdown export for completed dictations and meeting transcripts.
  • Fixed downloaded Whisper models sometimes appearing unavailable, especially before the online catalog loaded.
  • Improved multilingual transcription with persistent language settings and clear protection against English-only model mismatches.
  • Updated OpenAI, Anthropic, and Gemini recommendations; available models now refresh automatically without a manual refresh button.
  • Fixed formatting engine and model settings sometimes reverting after leaving Settings.

What's New in 1.2.3

  • Fixed Parakeet V3 failing because model downloads could omit a required vocabulary file.
  • Fixed Whisper returning phrases such as "Thank you." when no microphone audio was captured.

What's New in 1.1.3

  • Set minimum version to Sonoma.
  • Improve settings screen layout.

What's New in 1.1.1

  • New feature: Record meetings. NanoVoice can now record and transcribe your meetings locally, then turn them into clean, topic-organized notes.
  • A floating pill shows while a meeting is recording, with a live timer, audio meter, and one-click Stop.
  • New on-device LLM for much better formatting: download a built-in local model (no API key) for sharper notes, emails, and meeting summaries.
  • Redesigned Models settings: a clearer step-by-step layout for speech-to-text, formatting, and cloud keys.
  • You can now edit any built-in template to match how you write, with one-click restore to default.

What's New in 1.1.0

  • New feature: Record meetings. NanoVoice can now record and transcribe your meetings locally, then turn them into clean, topic-organized notes.
  • A floating pill shows while a meeting is recording, with a live timer, audio meter, and one-click Stop.
  • New on-device LLM for much better formatting: download a built-in local model (no API key) for sharper notes, emails, and meeting summaries.
  • Redesigned Models settings: a clearer step-by-step layout for speech-to-text, formatting, and cloud keys.
  • You can now edit any built-in template to match how you write, with one-click restore to default.

What's New in 1.0.3

  • Added local Parakeet V3 transcription using FluidAudio/CoreML.
  • Added in-flow model download prompts for Whisper and Parakeet when a required local model is missing.
  • Added a true raw Transcript template with no post-processing.
  • Moved the previous cleaned transcript behavior to Clean output for Lifetime users.
  • Improved microphone permission detection on newer macOS versions.

What's New in 1.0.1

  • Added more dictation language options, including Catalan, Swedish, Norwegian, Danish, Finnish, Greek, Czech, Romanian, Hungarian, Vietnamese, Thai, Indonesian, and Hebrew.
All NanoVoice releases on GitHub