Blurt vs Soniox

Side-by-side comparison of features, pricing, and ratings

Analysis reviewed Live tool data as of 2026-09-02
Cross-checked through our multi-step verification ·
Saved

At a glance

DimensionBlurtSoniox
PricingFree (open-source, no cloud dependency)Paid (token-based, ~$0.12/hour real-time; up to 10x cheaper than major cloud providers)
DeploymentLocal offline (GNOME Shell extension, whisper.cpp)Cloud API (real-time streaming, multi-language STT/TTS/translation)
LanguagesMultilingual (depends on whisper.cpp model chosen)60+ languages with code-switching support
Top FeatureFully offline, no account needed, privacy-firstSub-200ms streaming latency + real-time translation across 3,600 pairs
IntegrationsGNOME Shell onlyLiveKit, Tencent Cloud, Agora, Perplexity, etc.
ComplianceN/A (local execution)SOC 2 Type 2, ISO 27001, HIPAA, GDPR; audio never stored

Soniox and Blurt serve entirely different purposes: Soniox is a paid, cloud-based API for real-time multilingual STT, TTS, and translation with enterprise compliance, while Blurt is a free, offline GNOME extension for simple dictation on Linux. Choose Soniox if you need low-latency, multi-language voice agents or live translation at scale; choose Blurt if you are a Linux user who wants private, no-cost dictation in text fields without leaving the desktop.

Blurt
Blurt

Blurt: offline speech-to-text for GNOME with whisper.cpp

Visit Website
Soniox
Soniox

Multilingual speech AI API for real-time STT, TTS & translation

Visit Website
Pricing
Free
Paid
Plans
$0.10/hour
$0.12/hour
$0.70/hour
Popularity
2 views
7.2k views
Skill Level
Intermediate
Advanced
API Available
Platforms
Plugin
WebMobileDesktopAPI
Categories
🎤 Voice Dictation
Transcription & Speech-to-Text🎙️ Voice & Speech Translation & Localization
Features
Offline speech-to-text using whisper.cpp
Input speech into any GNOME text field
Multilingual speech recognition
Invoke via middle mouse button or CTRL+V
Runs entirely offline, no internet required
Lightweight C++ backend
Customizable whisper.cpp models
Supports GNOME Shell 42-48
Free and open-source
No account or cloud dependency
Requires `wsi` script installation
Active version 8 covers GNOME 46-48
Real-time speech-to-text streaming with sub-200ms latency
Async (batch) transcription at $0.10/hour
Text-to-speech generation in 60+ languages with expressive audio tags
Instant voice cloning from a few seconds of audio
Real-time speech translation across 3,600 language pairs
Multi-speaker diarization (bundled)
Language identification and code-switching support
Smart formatting and punctuation (bundled)
WebSocket and REST APIs for streaming and batch
SDKs for Python, Node, Web, React, React Native
In-region processing for data residency
Audio never stored—processed in memory
Compliance: SOC 2 Type 2, ISO 27001, HIPAA, GDPR
Soniox Compare tool to test STT, TTS, translation on your own data
Token-based pricing with no extra cost for diarization, translation, or formatting
Integrations
LiveKit
Pipecat
Agora
Tencent Cloud

What real users say: Blurt vs Soniox

Not marketing copy and not our opinion — a structured sweep of public discussion (reviews, forums, communities and video comments), showing what people praise and what they complain about for each tool.

Blurt

104 mentions across 6 sources · 22% positive — critical

Hacker News, Product Hunt, App Store, Bluesky, GitHub, Lemmy

What users praise

  • Offline speech-to-text with no internet required, ensuring privacy.
  • Works in any GNOME text field via middle-click or shortcut.
  • Lightweight C++ backend (whisper.cpp) keeps resource usage low.
  • Multilingual recognition thanks to Whisper model swapping.

What frustrates them

  • Setup requires manual installation of external wsi script.
  • Compatibility often broken on new GNOME Shell versions.
  • Error messages are cryptic and logging is poorly documented.
  • No built-in model downloader—users must source models themselves.

Researched Jul 4, 2026

Soniox

41 mentions across 2 sources · 80% positive

Hacker News, Bluesky

What users praise

  • Sub-200ms latency for real-time streaming.
  • Cost-effective pricing at 8-10x less than major cloud providers.
  • Multilingual support for 60+ languages with code-switching.
  • Bundled translation across 3,600 language pairs at no extra cost.

What frustrates them

  • Relatively expensive for low-volume or hobbyist use.
  • Requires API skills; no no-code integrations available.
  • Accuracy with heavy foreign accents can lag behind competitors.
  • Not available as a standalone macOS app or on App Store.

Researched Jul 16, 2026

Who should pick which

  • Enterprise developer building multilingual voice agents
    Pick: Soniox

    Soniox offers sub-200ms latency, real-time translation, multi-speaker diarization, and HIPAA/GDPR compliance — essential for customer support or sales agents handling global users.

  • Linux user wanting private offline dictation
    Pick: Blurt

    Blurt integrates directly into GNOME text fields, uses whisper.cpp locally with no internet, and is free — perfect for privacy-focused dictation on Linux.

  • Live event or meeting translator
    Pick: Soniox

    Soniox can stream translation across 3,600 language pairs in real-time, with language detection and code-switching, ideal for live multilingual events.

  • Hobbyist experimenting with STT on a budget
    Pick: Blurt

    Blurt is free and open-source, letting users test offline speech recognition without any cost or cloud dependency, using customizable whisper.cpp models.

  • Wearable or IoT developer needing low-latency STT
    Pick: Soniox

    Soniox's sub-200ms streaming and unified API for STT/TTS/translation fit low-power devices, with in-region processing for data residency.

Frequently Asked Questions

Blurt vs Soniox: which should you choose?

Soniox and Blurt serve entirely different purposes: Soniox is a paid, cloud-based API for real-time multilingual STT, TTS, and translation with enterprise compliance, while Blurt is a free, offline GNOME extension for simple dictation on Linux. Choose Soniox if you need low-latency, multi-language voice agents or live translation at scale; choose Blurt if you are a Linux user who wants private, no-cost dictation in text fields without leaving the desktop.

Is Blurt available for Windows or macOS?

No, Blurt is a GNOME Shell extension designed solely for Linux desktops running GNOME.

Does Soniox require any on-device processing?

No, Soniox is a cloud API. However, it offers in-region processing for data residency, and audio is never stored.

Can Blurt transcribe multiple speakers?

No, Blurt only transcribes speech input into a text field — it does not perform diarization or identify speakers.

Does Soniox have a free tier?

The provided data does not mention a free tier; Soniox is paid with token-based pricing.

Can I use Blurt for real-time translation?

No, Blurt is offline-only and provides speech-to-text without translation capabilities.

What compliance certifications does Blurt have?

Blurt is local and open-source, so no certifications apply; security depends on user's system.

Does Soniox support voice cloning?

Yes, Soniox includes voice cloning from just a few seconds of audio.

Which tool is better for a beginner?

Blurt is simpler to use if you have GNOME: install the extension and the `wsi` script. Soniox requires API integration, so it's better for developers.

More Blurt or Soniox comparisons

Explore each tool further

Browse these categories

Still deciding? Get the weekly AI tools brief

One email a week — new tools, honest comparisons, no spam.

Last reviewed: July 4, 2026