Voicemod is a desktop voice changer that routes your mic through a virtual audio device so you can speak as any of 200+ AI voices in Discord and in-game chat, paired with a soundboard and a Fairly Trained certification covering its voice models.
Used this tool? Rate it
Used this tool? Rate it
Voicemod is a real-time voice changer and soundboard for desktop computers, built primarily for people who talk while they play. Its homepage headline states the proposition in five words — "REAL-TIME VOICE CHANGER & SOUNDBOARD" — and the product has stayed unusually close to that description for years. You install it, pick a character voice, and everyone in your Discord call or game lobby hears that voice instead of yours, live, with no recording step in between.
The mechanism matters more than most product descriptions let on, because it explains both the reach and the limits of the tool. Voicemod installs what it calls a virtual microphone. After installation, you go into whatever application you want to speak in — Discord, a game, a meeting app — and select "Voicemod Virtual Microphone" as the input device instead of your actual headset. From that point on, the application has no idea it is not talking to a normal microphone. This is why the company can claim compatibility with more than fifty platforms without needing a partnership with any of them: it is not integrating with Discord or Fortnite, it is impersonating a microphone at the operating-system level.
The second half of the product is the soundboard, and it is not an afterthought. The soundboard lets you fire off audio clips into the same voice channel with a keypress, which in practice is how most of the meme culture around the tool actually happens. Alongside these sit Voicelab, a builder for making your own voices out of layered effects, and Voicemod Key, a piece of companion hardware that extends voice changing onto game consoles by pairing with a phone.
What separates Voicemod from most of the category is not a feature but a sourcing decision, and it is one that can be independently verified. The company states on its homepage that its AI voices are trained with professional voice actors and that its models are Fairly Trained certified. Fairly Trained is an independent certification body founded by former Stability AI executive Ed Newton-Rex, and its own published register lists Voicemod under company-level Licensed Model certification. Its March 2024 announcement identifies Voicemod as the first certified company offering AI speech and singing models, and explains that the certification is granted only where a company can demonstrate that training data was obtained under a contractual agreement with a party holding the necessary rights, or is public domain or openly licensed. The same announcement notes SAG-AFTRA and The Authors Guild joining as supporters of the scheme. In a category where training-data provenance is usually an unanswerable question, that is a substantive and checkable difference.
The company operates from Valencia, Spain. Its iOS application is published by an entity named Voicemod Inc., Sucursal en España, and the Valencia location is corroborated on its Trustpilot business profile.
Voicemod draws a distinction between two things that look similar from the outside, and it is worth understanding before choosing a voice. Traditional real-time filters process your existing voice — shifting pitch, adding distortion, layering a radio effect. AI voices, in the company's framing, rebuild the voice from scratch: the emotion, timing and energy of your delivery are preserved, but the person on the other end hears a different character entirely rather than a processed version of you. The practical consequence is that a filter tends to still sound like you underneath, while an AI voice tends not to.
Everything runs through the virtual microphone. Setup is three steps by the company's own description: install, which places the virtual microphone on your system; configure, where you select that device as the input in your target application; and use, where you pick a voice and talk. Because the transformation happens on your own machine and is presented downstream as an ordinary audio input, it works in applications that have never heard of Voicemod.
The company positions processing as local. Its own FAQ answers the safety question by stating that Voicemod runs on your device and processes your voice in real time, so you stay in control of how and where you use it. It advertises ultra-low latency alongside minimal resource impact on PC, mobile or console. There is corroborating evidence that the engineering is real rather than aspirational: the company has a partnership with Qualcomm under which, on Snapdragon X Elite hardware, its AI voice workloads run on the neural processing unit for lower latency, entirely on-device.
The company states it offers more than 200 Voicemod voices, spanning what it describes as everything from AI anime characters to Battlefield-style radio effects. Beyond the official set, community-made voices are browsable inside the application, so the effective library is considerably larger than the official count.
Voicelab is the creation tool, and it is closer to a modular effects rack than to a cloning system. The company describes over 140 voice effects that can be layered across three categories: Characters, Devices and Surroundings. In practice this means combining a personality with a transmission medium and an acoustic space — a character voice through a megaphone in a reverb-heavy room, for instance. You can build from scratch or remix an existing voice, then publish the result to the community for others to use and remix in turn.
Separately from the effects-layering in Voicelab, the terms of use describe a self-service tool called Voice AI Creator that trains a voice from uploaded recordings. This is the part of the product where consent becomes a legal question, and the terms address it directly. They state that you may only use the self-service creation tools when you have the rights to the inputs you are using, which means you can use your own voice or the voice of another person, provided that you have their consent or permission to use their voice and to use their name for the voice you will create.
Two details in that clause deserve attention. First, permission to use the voice and permission to use the person's name are treated as separate requirements, both of which you need — naming a voice after someone is not automatically covered by their agreement to be recorded. Second, the obligation is placed on the user rather than being verified by the platform; the terms make you responsible for the use you make of the tools and the content you produce with them. The acceptable-use rules reinforce this by prohibiting content that infringes any third party's copyright, trademark, privacy right or right of publicity, and content that is harassing, defamatory, false or intentionally misleading.
The soundboard lets you assign audio clips to keys and fire them instantly into a call. It is available in the free tier, where you can create a soundboard of your choice; paid users get unlimited soundboards. A Recorder feature captures clips directly into soundboards, with an instant replay function that rewinds up to thirty seconds.
Noise suppression and voice enhancement are built in, and this is the one paid-tier-adjacent capability the company keeps permanently available for free — its FAQ describes the free version as including a permanent clean mic effect to boost quality. The company's pitch here is that whatever your setup, the software makes the microphone sound better before any character voice is applied.
Voicemod Key is companion hardware. Pairing it with a smartphone unlocks voice changers and soundboards on a gaming console, which is the only route to using the product outside a desktop computer in real time.
The company lists five named partnerships, and they cluster tellingly around gaming peripherals. Elgato provides one-tap control over voice and soundboard switching. MSI's Mystic Light integrates so lighting reacts through Voicemod Sync. Razer has an exclusive voice collection unlockable through Razer Cortex. Corsair's iCUE pairs real-time voice filters with reactive lighting. Qualcomm handles the on-device AI acceleration described above.
Gaming voice chat. This is the primary case and the one the product is engineered around. The company maintains dedicated setup pages for individual titles — Valorant, Fortnite, Call of Duty, Roblox, GTA, Genshin Impact, Among Us, Sea of Thieves, Rust, Palworld and Escape from Tarkov among them — which means per-game quirks have generally already been documented.
Discord and community voice. Discord is both a usage venue and a distribution channel here. The company runs a Discord bot, maintains a dedicated soundboard-for-Discord page, and the community publishes Voicelab creations that others pull into their own setups.
Streaming. The company explicitly optimises for streamers alongside gamers, and the partner list makes the intent clear: Elgato and Corsair are streaming-desk staples, and the integrations tie voice switching to the same controls and lighting a streamer already runs.
Meetings and calls. Because the virtual microphone works in any application that accepts a microphone, the tool functions in Zoom, Microsoft Teams, Skype and similar. The company maintains a Microsoft Teams page and a video-call page. Worth remembering, though, that the terms describe the products as provided for leisure and entertainment purposes, which is a poor fit for anything load-bearing at work.
Voice acting and character work. A niche but real case, evidenced on the company's own review profile: one reviewer describes buying a lifetime licence in 2019 specifically as a voice actor, using the program to enhance certain performances when needed.
Privacy and persona. Speaking in public voice channels without exposing your own voice is a legitimate reason to use a changer, and the character-driven library serves it well.
Soundboard and meme culture. Distinct from voice changing entirely — many users are there for the ability to fire audio clips into a channel on a hotkey, not to alter their voice at all.
Feed it a clean signal. The company's own guidance is that AI voices track your speech more closely when the input is clean — a quiet room and a decent microphone, with noise removal on. This matters more with AI voices than with simple filters, because a rebuild has more to get wrong than a pitch shift does.
Set the input per application, not once globally. Each application keeps its own audio device setting. Working in Discord does not mean it is working in your game.
Test before it matters. Use the monitoring toggle and a low-stakes call to confirm both the voice and the latency before dropping into a competitive lobby or a live stream.
Treat the soundboard as a separate skill. Volume levels on community clips vary wildly. Check a clip's loudness before firing it into a channel, or you will deafen your squad.
Remember the terms frame the product as entertainment. The disclaimer describes the products as provided for leisure and entertainment purposes with no guarantee of suitability for a particular purpose. That is a reason to keep it out of anything consequential.
Understand what you grant when you publish. Uploading to the community hub keeps your ownership, but grants the company a perpetual, irrevocable, sublicensable licence covering storage, display, modification, derivative works, marketing use and sublicensing to third parties or APIs. If you would not want a creation used that way, keep it local.
Note the commercial-use carve-out before monetising. Content is licensed for personal, non-commercial use, except for monetisation generated by content creators through social networks and social media. If your use is a YouTube or Twitch channel, that exception is the clause that covers you; if it is a client deliverable, it likely is not.
Take advantage of the refund window if unsure. A thirty-day refund period applies to PRO purchases, handled through the support form, and the same thirty-day timeline is stated to apply to purchases made through distributors.
Be careful with lifetime licences. The terms state a lifetime licence remains active until you request account closure, but that once unsubscribed it cannot be reactivated and no refunds are given — you would need to buy a new licence.
Gamers who talk in voice chat are the core audience, and the per-title setup pages, low-latency engineering and peripheral partnerships all point the same direction.
Streamers and content creators are the explicitly named second audience. The Elgato and Corsair integrations fit an existing streaming desk, and the terms carve out social-media monetisation from the non-commercial restriction.
Discord communities get value beyond individual use, since soundboards and community-published voices are shared artefacts rather than personal settings.
People who want to mask their voice online are served by the character library without needing any of the creation tools.
Anyone who cares about training-data provenance has an unusual amount to work with here. If you have reservations about AI voices trained on scraped material, the Fairly Trained certification is verifiable at the certifier's own register rather than resting on the vendor's word.
Who should look elsewhere: anyone who needs real-time voice changing on a phone, since that is explicitly not what the mobile app does; anyone needing a produced voiceover file rather than live transformation, where text-to-speech and voiceover tools fit better; anyone requiring the tool for professional or business-critical work, given the entertainment-purposes disclaimer; and anyone who needs published pricing before installing, since the company does not publish a pricing page.
Desktop, Windows and Mac. This is the only place real-time voice changing works. The virtual microphone architecture is a desktop operating-system feature, and every capability described above depends on it.
Mobile, iOS and Android. The mobile app is a soundboard and recorded-message tool, not a live voice changer. The company states this outright in its App Store description: real-time voice use is desktop only, and the app does not work live in voice chat apps or games. What it does offer is a library the company describes as over 500,000 meme sounds, unlimited custom soundboards, and the ability to record voice messages with voice effects and share them. On Apple's own listing data, the iOS app holds a 4.7 rating across 19,693 ratings, carries a 17+ content rating, is free, requires iOS 16.0 or later, first shipped in December 2023 and was updated to version 1.23.19 in August 2026.
Consoles. Reachable only through Voicemod Key, the companion hardware that pairs with a smartphone.
Applications. The company claims compatibility with more than fifty platforms, covering both dedicated voice apps like Discord and in-game voice chat, with named setup pages for TeamSpeak, Skype, Microsoft Teams and a long list of games.
The product operates on a free tier plus a paid tier called Voicemod PRO.
What the free tier includes, per the company's own FAQ: a rotation of voice changing options rather than the full library, a permanent clean mic effect for audio quality, and the ability to create a soundboard. The company frames this as sufficient for gaming, streaming or chatting with friends.
What PRO adds: access to more voices and sounds, full customisation options, and unlimited soundboards.
A note on price figures, and why this page does not give any. Voicemod does not publish a pricing page. Checked directly, the common paths all fail: several obvious pricing URLs return 404, one redirects to a broken target, and the company's own page sitemap — 353 pages — contains no pricing page at all. Third-party sites quote figures that contradict each other substantially, and none of them can be reconciled against an official source. Rather than repeat numbers that cannot be verified, the honest answer is that current pricing is visible inside the application or at official distribution channels, and that is where you should check it.
Licence structure is documented in the terms, even though prices are not. PRO is available both as a recurring subscription and as a lifetime licence — a one-time payment described in the terms as OTL, "you pay once and lasts forever." Recurring subscriptions can be cancelled any time before the next renewal period by contacting support. A lifetime licence stays active until you request account closure, but the terms warn that once unsubscribed it cannot be reactivated and no refunds are given; obtaining PRO again would require a new licence.
Refunds. Thirty calendar days from purchase if you are not satisfied, handled by the support team through the support form. Purchases made through distributors are subject to those distributors' own procedures, though the terms state the thirty-day timeline still applies.
Device limits. PRO content can be accessed on multiple devices you sign in to, but used on only one device at a time.
Voice.ai is the closest direct comparison on real-time voice changing, and the two appear side by side in Trustpilot's related-companies panel. It has moved toward a broader platform including text to speech and telephone voice agents, whereas Voicemod has stayed focused on live transformation for gaming and streaming.
Clownfish Voice Changer is the long-standing free, lightweight alternative in this space. Its Trustpilot profile carries too few reviews to support any quality conclusion, so it is worth evaluating hands-on rather than by rating.
Text-to-speech and voiceover platforms such as ElevenLabs solve a different problem. If you need a produced audio file with a synthetic voice, a voice changer is the wrong tool regardless of quality — nothing in Voicemod is built around generating narration from text.
Built-in platform effects. Discord and various games ship their own voice modulation. These cost nothing and require no virtual audio routing, but the range is far narrower.
Hardware voice changers and DSP mixers. The classical approach, still used by some streamers. The company notes many of its users arrive from exactly these tools.
Real-time voice changing does not work on mobile. The single most important limit, and the one most likely to disappoint someone arriving from the marketing. The mobile app is a soundboard and recorded-message tool; the company documents this clearly in its App Store listing.
There is no published pricing. No pricing page exists on the site, one pricing URL redirects to a broken target, and the sitemap has no pricing entry. You cannot compare cost against alternatives without installing or going to a distributor, and third-party price quotes conflict.
Community review scores are low and internally inconsistent. Trustpilot shows a TrustScore of 1.7 across 456 reviews. But the star distribution on that same page is 55% five-star, 12% four-star, 3% three-star, 6% two-star and 24% one-star — a simple weighted average of which lands near 3.9, not 1.7. Trustpilot's score is time-weighted rather than a plain average, and the page notes only 44 reviews in the last twelve months, which is the likely explanation. Both figures are reported here rather than reconciled. Two further caveats: the profile is flagged as having no recent history of asking for reviews, which Trustpilot itself says may make reviews unrepresentative, and the company has replied to 62% of negative reviews. Set against this, the iOS app holds 4.7 across 19,693 ratings — a sample forty times larger and dramatically more positive. These channels disagree, and both are reported.
What the complaints are actually about. Trustpilot's own topic labels for this profile cluster around Customer communications, Order, Customer service, Delivery service and Refund — that is, commercial and support friction rather than the voice changing itself.
Content is licensed for personal, non-commercial use. With the notable exception of creator monetisation through social networks and social media. Anything outside that carve-out falls outside the licence.
You grant a broad licence when you upload. Ownership stays with you, which is better than several competitors offer, but the licence you grant is perpetual, irrevocable, sublicensable and covers derivative works, marketing use and sublicensing to third parties or APIs.
The product is disclaimed as entertainment. The terms state it is provided as-is and as-available, for leisure and entertainment purposes, without guarantee of suitability for a particular purpose.
Age restrictions differ across sources. The terms set a minimum of 16, with parental agreement required for minors over 16. Apple's listing rates the iOS app 17+. These are different thresholds from different bodies and are not reconciled anywhere.
Lifetime licences are not recoverable once cancelled. Unsubscribing ends it permanently, with no refund and no reactivation.
Consent for cloned voices is your responsibility. The terms require rights to your inputs, but the burden of verifying them sits with you, and the indemnification clause makes you liable for third-party claims.
Setup can fail per application. Because each application has its own input device setting, working in one place does not mean working everywhere. This is the most common practical friction with the virtual microphone approach.
Yes, with real limits. The company states the free version includes a rotation of voice changing options, a permanent clean mic effect to boost quality, and the ability to create a soundboard, and describes this as suitable for gaming, streaming or chatting with friends. Upgrading to PRO adds more voices and sounds, full customisation and unlimited soundboards. The free tier is genuinely usable rather than a time-limited trial, but the voice selection rotates instead of being fully open.
This cannot be answered from official sources, and it is worth explaining why rather than guessing. Voicemod publishes no pricing page: the obvious URLs return 404, one redirects to a broken target, and the company's own sitemap of 353 pages contains no pricing page. Third-party sites quote figures that contradict each other. What the terms do document is the licence structure — PRO is sold both as a recurring subscription and as a one-time lifetime licence — along with a thirty-day refund window. For current prices, check inside the application or at an official distribution channel.
Yes. The terms state you may only use the self-service creation tools when you have the rights to the inputs you are using. Concretely: you can use your own voice, or another person's voice provided you have their consent or permission to use their voice and to use their name for the voice you create. Note that those are two separate permissions — using the voice and naming the voice after them. The terms also prohibit content infringing any third party's privacy right or right of publicity, and you are responsible for the content you produce and liable for third-party claims arising from it.
This is one of the few cases in the category where the claim is externally checkable. The company states its AI voices are trained with professional voice actors and that its models are Fairly Trained certified. That certification is verifiable at the certifier's own site: Fairly Trained's published register lists Voicemod under company-level Licensed Model certification, and its March 2024 announcement identifies Voicemod as the first certified company offering AI speech and singing models. The certification is granted only where training data was obtained under a contractual agreement with a rights-holding party, or is public domain or openly licensed. Fairly Trained was founded by former Stability AI executive Ed Newton-Rex, and SAG-AFTRA and The Authors Guild are among its supporters.
Not for real-time voice changing. The mobile app is a separate product for soundboards and recorded voice messages, and the company states directly in its App Store listing that real-time voice use is desktop only and the app does not work live in voice chat apps or games. For live transformation you need the Windows or Mac desktop application. The mobile app does offer a large meme sound library, unlimited custom soundboards, and recording voice-changed messages to share.
The company claims more than fifty platforms, covering both dedicated voice applications like Discord and in-game voice chat. The mechanism is generic rather than per-app: because it presents itself as a virtual microphone, it works in essentially anything that accepts a microphone input. The company maintains individual setup pages for Valorant, Fortnite, Call of Duty, Roblox, GTA, Genshin Impact, Among Us, Sea of Thieves, TeamSpeak, Skype and Microsoft Teams among others.
Only within a specific carve-out. The terms license content for your personal, non-commercial use, except for monetisation generated by content creators through social networks and social media. That exception covers a monetised YouTube or Twitch channel. It does not obviously extend to client work, advertising or commercial production, and the terms separately disclaim any warranty of suitability for a particular purpose. Get clarity in writing before building commercial deliverables on it.
The company addresses this directly, stating it will not be listening to sound chats nor your voice recordings, and that it needs microphone access only to process your voice in order to apply the voice filters you choose. It also describes processing as happening on your own device in real time. Separately from listening, note the licence you grant over content you choose to upload to the community hub, which is broad.
Because the two are measuring different populations, and one of them is internally odd. Trustpilot shows 1.7 across 456 reviews, while Apple shows 4.7 across 19,693 — a sample forty times larger. Complicating this further, Trustpilot's own star distribution for the profile is 55% five-star against 24% one-star, which averages closer to 3.9 than to 1.7; the score is time-weighted rather than a simple average, and only 44 reviews arrived in the last twelve months. Trustpilot also flags the profile as having no recent history of asking for reviews, which it says may make reviews unrepresentative. The topic labels on the negative reviews centre on customer service, orders and refunds rather than the software's core function. The honest reading is that the app experience and the commercial-support experience are rated very differently.
It ends permanently. The terms state that a lifetime licence remains active until you request the closure of your account, but that once you have been unsubscribed it cannot be reactivated and no refunds will be given, and that obtaining PRO again would require purchasing a new licence. This is worth knowing before closing an account you might want back.
