Sotto vs Hearsy: Local Mac Dictation Compared

Sotto and Hearsy occupy the same narrow niche in the Mac dictation market: both run Parakeet and Whisper locally, both require a one-time payment, and both auto-paste transcribed text wherever your cursor is. If you're deciding between them, the differences are smaller than the similarities — which makes them worth examining carefully.

One disclosure upfront: Hearsy is my product. I've used Sotto to write this comparison and have tried to be accurate about where it does things differently or better.

---

What Sotto is#

Sotto is a macOS dictation app that runs speech-to-text locally using WhisperKit and Parakeet. You hold a hotkey, speak, and release to insert text. It works system-wide across any Mac app.

As of March 2026, Sotto costs $49 as a one-time purchase covering up to 3 Macs with lifetime updates. There's no subscription.

Transcription engines available in Sotto:

• WhisperKit variants: Tiny, Base, Small, Large V3 Turbo, Distil Large V3 — all running locally via Apple's Neural Engine
• Parakeet v2 — English, local
• Parakeet v3 — Multilingual, local
• Cloud options (optional): OpenAI gpt-4o-mini-transcribe ($0.006/min) or Groq Whisper Large V3 Turbo ($0.006/min) — these require API keys and an internet connection

Sotto also includes:

• Custom vocabulary dictionary for specialized terms
• Automatic text cleanup rules (local, pattern-based — filler word removal, grammar fixes)
• AI-powered functions for email drafting, code prompt optimization, and translation — these use cloud APIs (OpenAI or Groq)
• Audio file import and transcription
• Recording history with re-transcription using any available model
• Per-language hotkey assignment
• 90+ language support

What Sotto is: A local Mac dictation app with WhisperKit and Parakeet models, optional cloud transcription, a custom vocabulary dictionary, and cloud-based AI functions. One-time $49 purchase, 3-Mac license.

---

What Hearsy is#

Hearsy is a macOS menu-bar dictation app. Press a global hotkey from any app, speak, and text is pasted at your cursor. It supports two transcription engines:

• Parakeet TDT — NVIDIA's model, optimized for English, runs in under 50ms on Apple Silicon
• Whisper Large — OpenAI's model, 99 languages, 1–2 second latency

Where Hearsy differs most from Sotto is in the AI cleanup layer. Hearsy ships with Qwen 2.5 3B via MLX — a local language model that runs entirely on your Mac. When you apply a template (Clean & Format, Email, Code Comment, Summary), the processing happens on-device with no API key and no external call. You can optionally configure Claude or OpenAI as the cleanup provider, but the full AI pipeline is available locally from the moment you install the app.

Hearsy is a one-time purchase. No subscription, no per-minute charges.

What Hearsy is: A macOS dictation app with Parakeet and Whisper engines, an on-device LLM (Qwen 2.5) for AI cleanup, and structured AI templates. One-time purchase. macOS only.

---

At a glance#

---

Transcription engines: similar on paper, different implementations#

Both apps support Parakeet, which means English latency is similar — under 50ms on Apple Silicon. For most English-speaking users, raw transcription speed will feel the same between the two apps.

The difference is in how they run the Whisper variants. Sotto uses WhisperKit, a Swift library developed by Argmax that runs Whisper models via Apple's Neural Engine and Core ML. WhisperKit is optimized for the Neural Engine, which can be more power-efficient and leaves the GPU available for other work.

Hearsy uses whisper.cpp for its Whisper engine, which runs on the GPU or CPU. Performance is comparable on Apple Silicon, but the hardware paths differ. Sotto's Neural Engine optimization may have an edge on sustained battery life for users who dictate extensively on a laptop.

For non-English transcription, both apps use Whisper — Sotto via WhisperKit, Hearsy via whisper.cpp. Accuracy should be similar since both reach for comparable Whisper model sizes.

Sotto also offers a wider selection of local Whisper model sizes (Tiny through Distil Large V3), which is useful if you want to trade accuracy for lower memory usage on older hardware.

---

AI cleanup: local LLM vs local rules vs cloud API#

This is the most substantive difference between the two apps.

Sotto's AI enhancement:

Sotto has two layers of text processing:

Automatic cleanup rules are local and pattern-based. The app can remove filler words ("um," "uh") and apply grammar fixes without making an API call. This works offline.

AI-powered functions — email drafting, code prompt optimization, translation — use cloud APIs (OpenAI or Groq). You supply your own API keys. These are more powerful than cleanup rules but require internet access and incur per-request costs at their published API rates (~$0.006/min for Groq).

Sotto does not include a local LLM. There's no on-device model for contextual rewriting.

Hearsy's AI enhancement:

Hearsy ships with Qwen 2.5 3B via MLX, which runs locally. Applying a template (Clean & Format, Email, Code Comment, Summary) runs the LLM on your Mac — nothing leaves your device. The local model handles contextual rewriting, not just pattern-based cleanup: it understands sentence structure, reformats prose, and generates structured output like email bodies or formatted code comments.

You can also configure Claude or OpenAI as the cleanup provider in Hearsy if you prefer cloud model quality, but it's opt-in. The local LLM is available immediately with no configuration.

The practical comparison:

For users dictating sensitive content — medical notes, legal documents, confidential business communication — Hearsy's full pipeline (transcription plus AI formatting) stays on-device without any cloud configuration. Sotto's cleanup rules work offline, but any AI-assisted rewriting requires a cloud API call.

For users who primarily want filler word removal and light grammar correction, Sotto's local cleanup rules handle that without a cloud dependency. The AI function layer in Sotto is more capable for specific use cases like email drafting, but it's cloud-only.

---

Pricing and licensing#

Sotto (as of March 2026): $49 one-time, 3 Macs, lifetime updates. No subscription.

Hearsy: One-time purchase, single Mac. No subscription.

Sotto's multi-Mac licensing is a real advantage for users who work across machines. A $49 license covering a primary Mac, a MacBook for travel, and a second personal machine is straightforward value. Hearsy's one-time price applies to a single device.

If you have one Mac, the pricing comparison is direct. If you need coverage across multiple Macs, Sotto's 3-seat license is meaningfully different.

---

Features Sotto has that Hearsy doesn't#

Multi-Mac licensing. Three devices for one purchase. This is Sotto's clearest structural advantage.

Audio file import. Sotto can transcribe recordings, video files, and audio uploads — not just live dictation. If you record meetings or voice memos and need them transcribed, this is a significant capability Hearsy doesn't have.

Custom vocabulary dictionary. Sotto lets you add specialized terms — product names, medical terminology, technical jargon — that the model should recognize accurately. Hearsy doesn't have a user-editable vocabulary layer.

Re-transcription. Sotto saves recordings and lets you re-transcribe them with a different model (e.g., switch from Tiny to Large V3 Turbo after the fact). Useful if you recorded something quickly with a small model and want higher-quality output later.

Wider Whisper model selection. Sotto offers more granular control over which Whisper model to use, from Tiny (fastest, lower accuracy) to Distil Large V3 (good accuracy, smaller than full Large).

---

Features Hearsy has that Sotto doesn't#

On-device LLM for AI cleanup. Hearsy's Qwen 2.5 3B model handles contextual rewriting on your Mac — no API key, no internet, no per-request cost. This includes formatting, grammar correction, and structured output templates. Sotto's AI functions are cloud-only.

Structured AI templates. Clean & Format, Email, Code Comment, Summary — applied consistently each time. If you regularly dictate emails or code documentation, these templates apply predictable processing without needing to craft prompts.

Full offline AI pipeline. Transcription and AI formatting both run locally in Hearsy. Sotto's local layer covers transcription and pattern-based cleanup; contextual AI enhancement requires cloud APIs.

---

Sotto vs Hearsy vs SuperWhisper#

SuperWhisper is commonly compared alongside both apps. The key difference: SuperWhisper uses Whisper models only, with no Parakeet support. English latency is 1–2 seconds for most Whisper configurations. SuperWhisper has an established user base and a free tier with limited capabilities.

For a detailed three-way comparison covering Wispr Flow, SuperWhisper, and Hearsy, see Wispr Flow vs SuperWhisper vs Hearsy. For a broader look at local dictation options, see best Whisper apps for Mac.

---

Which to choose#

Choose Sotto if:

• You need to license one purchase across multiple Macs — 3-Mac coverage at $49 is a real advantage
• You transcribe audio files and recordings, not just live dictation
• You want a custom vocabulary dictionary for domain-specific terminology
• Local pattern-based cleanup (filler word removal) is sufficient for your AI needs, or you're comfortable supplying OpenAI/Groq keys for advanced AI functions
• You want granular control over Whisper model size

Choose Hearsy if:

• You want fully local AI enhancement — transcription plus contextual cleanup — with no API keys or cloud calls
• You dictate sensitive content and need the entire pipeline (transcription and AI formatting) to stay on-device
• Structured AI templates (Email, Code Comment, Summary, Clean & Format) match your workflow
• You're a single-Mac user and the multi-device licensing difference doesn't apply

Choose macOS built-in dictation if:

• Your dictation is occasional and under 30 seconds at a time
• You don't want to install third-party software

---

For more on local Mac dictation options, see best dictation software for Mac. For how Parakeet compares to Whisper technically, see Whisper vs Parakeet. For voice data privacy implications, see the voice data privacy guide. For a broader comparison of alternatives including Spokenly and BetterDictation, see best privacy-first dictation apps.

---

Frequently asked questions#

Does Sotto work offline?#

Yes. Sotto's local models — WhisperKit variants and Parakeet v2/v3 — run entirely on your Mac with no internet connection required. Cloud transcription (via OpenAI or Groq API keys) requires internet, as do AI-powered functions like email drafting and translation. Core push-to-talk dictation works fully offline using local models.

How many Macs does Sotto work on?#

Sotto's $49 one-time purchase covers up to 3 Macs with lifetime updates. Most Mac dictation apps — including Hearsy — license to a single machine. If you work across multiple Macs, Sotto's licensing structure is a material advantage.

Does Sotto use Parakeet?#

Yes. Sotto supports NVIDIA Parakeet v2 for English transcription and Parakeet v3 for multilingual transcription, alongside WhisperKit variants. Both Sotto and Hearsy support Parakeet, which delivers under-50ms English latency on Apple Silicon. The differences between the apps are in AI cleanup, licensing, and auxiliary features rather than raw transcription speed.

What is the difference between Sotto and Hearsy?#

The main differences: Sotto licenses to 3 Macs ($49), includes audio file import, and offers a custom vocabulary dictionary. Hearsy includes a local LLM (Qwen 2.5 3B via MLX) for contextual AI cleanup that runs on-device with no API key — Sotto's AI functions use cloud APIs. Both apps use Parakeet and Whisper for local transcription.

What is the best Sotto alternative for Mac?#

Hearsy is the closest alternative — same Parakeet and Whisper engines, local processing, one-time purchase, with a local LLM for AI cleanup. SuperWhisper uses Whisper only (no Parakeet), so English latency is higher. BetterDictation uses Whisper via Apple's Neural Engine. For a full list, see best Whisper apps for Mac.