CrystalSound
CrystalSound is a desktop app for Windows, macOS and Linux that removes background noise from both sides of a call, isolates your voice alone, and records screen and audio, with all processing happening on your own machine.
What is CrystalSound?
CrystalSound is a desktop application for Windows, macOS and Linux that cleans up the audio of online meetings using AI, and it does so without sending any sound off your machine. It is published by Nami Technology JSC, a Vietnamese company that also sells an enterprise voice AI stack, and CrystalSound is the speech-enhancement layer of that stack sold as a standalone product. Rather than plugging into one particular meeting platform, the app installs itself as a virtual microphone and a virtual speaker on your computer. You then pick CrystalSound as your input or output device inside whatever you already use, which is why the homepage shows icons for Google Meet, Microsoft Teams, Skype, Slack, Discord, Loom and OBS, and why a customer quote also mentions Zoom. Nothing has to be installed on the other side of the call. The core of the product is bi-directional noise cancellation: it strips background noise out of your own microphone and out of the audio you receive, so a noisy colleague on the far end is handled too. On top of that sits My Voice Only, which asks you to record a voice profile and then lets through that voice alone, muting the people talking around you. Both the audio and the voice profile are processed and stored locally; the publisher states it has no access to either. The app also handles acoustic echo suppression, room echo removal, residual noise removal, microphone volume boosting, 48 kHz high-definition stereo and a power-efficient mode, plus voice effects such as pitch shift, reverb and a robotic filter. Beyond live calls, CrystalSound records your screen together with both sides of the conversation, and can improve existing mp3, wav and flac files. Meeting minutes, automatic titles and keyword or sentiment analysis are advertised prominently, but the pricing grid currently marks them as unavailable on the Premium plan, and transcription belongs to an Elite tier that is still listed as coming soon. Team administrators can assign and revoke Premium licences and resize their number of seats at any time.
What it does
- Remove background noise on both ends of a call
- Isolate your voice alone and mute the people around you
- Record your screen together with both sides of the conversation
- Clean up existing mp3, wav and flac recordings
- Cancel acoustic echo, room echo and residual noise
- Boost and even out your microphone level
- Assign and revoke Premium licences across a team
When to use CrystalSound / When not to
A quick filter to help you decide if CrystalSound is the right fit.
When to use CrystalSound
- Anyone joining video calls from a noisy home, café or open-plan office
- Privacy-conscious teams that cannot let meeting audio leave the device
- People who switch between several calling apps, since the tool appears as a virtual microphone in all of them
- Linux users, who are rarely served by noise-cancellation software
- Team leads who need to assign, revoke and resize Premium seats themselves
When not to use CrystalSound
- Mobile and tablet users, as there is no iOS or Android application
- Developers looking for an API, since none is published
- Buyers who need meeting minutes, automatic titles or sentiment analysis today, as the pricing grid marks them unavailable on Premium
- Owners of older hardware, below an Intel 5th-generation Core i3 or an Apple M1, or still on macOS 10.15
- Anyone who cannot install software with administrator rights, as there is no browser version
How to use CrystalSound
A typical end-to-end flow, from setup to results.
- Download the package for your operating system from the CrystalSound download page
- On macOS, open the package, grant administrator rights, then allow microphone access
- On Windows, run the installer and tick Run CrystalSound MVO on the final screen
- On Linux, make the AppImage executable and install the fuse and libfuse2 dependencies
- Sign in with Google, with Microsoft, or with a login link sent to your email address
- Record a voice profile from Register Profile, speaking clearly in a quiet room
- Keep up to five profiles and rename, update or delete them from Manage Profile
- Select CrystalSound as the microphone and speaker inside your calling application
- Behind a corporate proxy, whitelist the seven authentication domains listed in the installation guide
- Follow the tutorials on the @namitech YouTube channel or the PDF user manual
Pros & Cons
Pros
- Audio and voice profile are processed on the device and never sent out
- Works with any calling app, since it registers as a virtual microphone and speaker
- Noise cancellation applies to the other party as well, not only to your own microphone
- Linux is supported alongside Windows and macOS, which is uncommon in this category
- The seven-day Premium trial ends on the free Starter plan with no charge
- Purchasing-power-parity pricing makes the tool markedly cheaper outside the United States
- A 30-day money-back guarantee is advertised on the pricing page
Cons
- No mobile or tablet application at all
- The meeting-intelligence features promoted on the homepage are marked unavailable on Premium
- The Premium price is injected by JavaScript and absent from the pricing page itself
- The blog and What's new pages were both empty when this review was written
- No public API and no developer documentation
- No DPA, no Article 27 EU representative and no explicit GDPR compliance claim
- The pricing page advertises a 30-day refund while the terms carry a non-refund clause
Pricing & Plans
A permanent free Starter plan is available, and the seven-day Premium trial reverts to it automatically without any charge. The cheapest paid entry point is the Premium plan at 12.00 USD per user per month, or 84.00 USD per user per year. The Elite plan is announced at 29 USD per user per month and 228 USD per user per year but has not been released yet, and Enterprise pricing is available on request only. Prices are adjusted to local purchasing power outside the United States, so the amount displayed varies by country.
- free of charge
- and the plan accounts fall back to when the Premium trial ends
- 12.00 USD per user per month or 84.00 USD per user per year
- for professionals and teams
- with unlimited noise reduction and cloud call recording in preview
- 29 USD per user per month or 228 USD per user per year
- listed as coming soon
- adding meeting transcription
- performance analysis
- risk analysis and areas of improvement
- price on request
- for large privacy-focused teams
- with secure offline screen recording
- bulk order discounts and virtual desktop infrastructure support
Data, GDPR & hosting
A consolidated view of how CrystalSound handles your data.
GDPR overview
The words GDPR and General Data Protection Regulation appear nowhere on the site, so there is no explicit compliance claim to report, and nothing denying it either. Some GDPR vocabulary is nevertheless used: the privacy statement distinguishes between acting as a data controller and as a data processor, and sets a minimum age of 16, which matches Article 8. Beyond that the framework is thin. No Article 27 EU representative is designated, no Data Processing Agreement is published or offered on request, no Data Protection Officer is named, and no transfer mechanism such as standard contractual clauses is mentioned. The publisher is established in Vietnam and has no EU entity. The data-protection contact is a personal address, anh@crystalsound.ai. The privacy statement has been in force since 20 June 2023.
Who owns the data?
Your audio never leaves your computer. CrystalSound processes speech and your VoiceID locally, and the publisher states plainly that it has access to neither. The voice profile you record for My Voice Only is stored on your own device, is never collected, and is never shared with third parties. Nami Technology JSC acts as a data controller for site visitors and customers, and as a data processor when it handles end-user data on behalf of its business clients. Payment details are never collected directly, being handled by PCI-compliant third-party processors. Account and personal data deletion is available on request to support@crystalsound.ai, and is permanent and irreversible.
Reuse rights
What the publisher does collect is usage telemetry: app version, current settings, email address, operating system, session start and end times, and which applications used the virtual microphone or speaker. Paying users can switch this tracking off in the app settings, which free users cannot do. Reporting a problem from inside the app uploads two log files containing operational logs, the names of your audio devices, your email address and your settings. The website uses Google Analytics cookies and referral-tracking cookies for its invite-a-friend feature. Data may be shared with technical service providers, with regulators where legally required, and with a buyer in the event of a merger or acquisition. The publisher explicitly rules out using your voice profile for AI training, directly or indirectly.
Data retention & training
Hosting summary
There is no hosting jurisdiction for the audio itself, because the audio is never uploaded: noise cancellation and voice isolation run on the user's own computer, and the voice profile is stored there too. What is hosted is the surrounding service. The website and backend run on Vercel, and the publisher uses four Google Firebase services: Cloud Firestore, Cloud Functions for Firebase, Cloud Storage for Firebase and Firebase Authentication. No country or region of hosting is named anywhere in the privacy statement. One technical clue points to the United States, the Cloud Functions domain being us-central1-crystalsound-mvo.cloudfunctions.net, but the publisher never states this as a commitment and it should not be read as one. IP addresses are logged by Firebase Authentication for a few weeks and then removed automatically. The ISO 27001, ISO 27017, ISO 27018 and SOC 1, 2 and 3 certifications cited in the privacy statement are those of Vercel and Firebase, not of CrystalSound.
Things to keep in mind
Risks and trade-offs to weigh before adopting CrystalSound.
- Recording a meeting may legally require the consent of everyone on the call, and the tool does not handle that obligation for you
- The homepage sells meeting minutes and sentiment analysis that the pricing grid marks unavailable on Premium, so check what you are actually buying
- The pricing page promises a 30-day refund while the terms of use contain a non-refund clause
- Subscriptions renew automatically unless you cancel before the renewal date, and fees already charged are not refundable
- The ISO 27001 and SOC certifications quoted in the privacy statement belong to Vercel and Firebase, not to CrystalSound itself
- Your voice profile is a biometric identifier; it stays on your device, but losing control of that device matters more than usual
- The publisher may amend the terms of use with immediate effect as soon as they are posted
Setup & Integrations
Technical difficulty
Moderate, and it depends on your operating system. Windows and macOS use ordinary installers, though both need administrator rights and macOS also asks for microphone permission. Linux is more demanding: the AppImage must be made executable and the fuse and libfuse2 packages installed. One step is specific to this product, recording a voice profile in a quiet room to enable My Voice Only. Nothing needs configuring in your calling app beyond selecting the virtual microphone. Behind a corporate proxy, seven domains must be whitelisted.
Deployment
Integrations
Supported languages
Behind CrystalSound
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Will I be charged when the Premium trial ends?
Is my audio uploaded to a server?
Which operating systems are supported?
Is there a mobile version?
Which meeting apps does it work with?
How much does Premium cost?
Is my data used to train AI models?
How do I delete my account and my data?
What is the minimum age to use CrystalSound?
Is there an affiliate programme?
Should you pick CrystalSound?
CrystalSound answers a narrow problem with unusual conviction. Noise cancellation and voice isolation run on your own computer, the voice profile never leaves the device, and the publisher states plainly that it does not use it to train models. For anyone taking calls from a shared flat, a café or an open-plan office, and especially for teams that cannot let meeting audio reach a third-party server, that architecture is the whole argument, and it is a good one. Working as a virtual microphone rather than a per-platform plugin is the other sensible decision, and Linux support is genuinely rare in this category. The reservations are about what surrounds the audio engine. The homepage sells meeting intelligence, minutes, keywords, sentiment, that the pricing grid marks unavailable on Premium, with transcription pushed to a plan that has not shipped. A buyer who signs up for the meeting-notes promise will not find it. The pricing page advertises a 30-day money-back guarantee while the terms of use carry a non-refund clause, a contradiction worth clarifying before paying. There is no mobile app, no public API, and the blog and release-notes pages were both empty when this review was written. On compliance the picture is thin rather than alarming: the site never claims GDPR compliance, publishes no DPA, names no EU representative, and the ISO and SOC certifications quoted in its privacy statement belong to Vercel and Firebase, not to CrystalSound. At 12.00 USD per user per month, with a free Starter plan and prices adjusted to local purchasing power, the entry cost is low enough that the audio engine can be judged on its own merits, which is where this product is strongest.
- Choosing a selection results in a full page refresh.
- Opens in a new window.