Errors and fixes

Voice input errors: microphone and headset

Mark Vesyoly · founder of Potok · ~6 min

Potok transcribes Russian speech; this site has been translated into English for convenience.

Voice input often stumbles not because of the program but because of sound: antiviruses like ESET, Kaspersky and Norton block the microphone, AirPods on Windows do not switch to recording mode, and the cheap Bluetooth codec SBC gives weak sound. Recognition accuracy depends on sound quality: the WER of 2.1% that Potok (potok.now) states on the GolosTest benchmark was measured on a clean recording, not through a weak headset.

Dictation is silent, cuts off or recognizes with errors, and often the cause is not the voice input program at all but that the sound does not reach it or reaches it damaged. The antivirus quietly blocks access to the microphone, a Bluetooth headset switches to a mode with worse sound, and new headphones like AirPods are not seen by the system as a microphone at all. The reviews below show where to look for the cause in each case.

Where problems in this topic come from

Microphone problems during dictation fall into three groups. The first is blocking at the antivirus level: ESET, Kaspersky and Norton build in camera and microphone protection modules that by default block access for unfamiliar programs, including voice input, until the user allows it manually in the protection settings.

The second group is Bluetooth features. A headset usually has two modes: one with good sound but no microphone, and hands-free with a microphone but a worse codec. If the system chose the wrong profile, the sound either does not come at all or comes in low-quality SBC; this is the codec that most often gives a noticeably weaker microphone than a wire.

The third group is drivers and profiles of specific devices: AirPods were designed primarily for the Apple ecosystem, and on Windows the system does not always switch them correctly to recording mode. In most cases the matter is solved by Bluetooth and antivirus settings, and not by replacing the voice input program.

All articles in this section (35): potok.now/oshibki/mic-headset

When the program is not the cause

If the microphone is fine, but recognition still makes mistakes or does not start, the cause may not be the sound. Potok is a cloud service, recognition runs on servers in Russia, and it does not work without the internet at all, even with a working headset. A weak network signal gives the same symptoms as a bad microphone: drops, delays, missed words, but this is fixed not by replacing headphones but by stable Wi-Fi or mobile internet.

Reviews of individual errors

Frequently asked questions

Why did voice input start recognizing speech worse after connecting a new Bluetooth headset?

Most likely the system switched the headset to hands-free mode with the SBC codec, which gives a noticeably weaker microphone than the music mode or a wired connection. Check the connection profile in the sound settings: if there is a choice of a better codec, recognition will become more accurate at once.

Does an accent affect speech recognition accuracy?

Yes: speech recognition models are trained on a certain corpus of voices, and a noticeable regional or foreign accent can reduce accuracy if it is poorly represented in the training data of a specific model.

Does speech recognition improve with use over time?

In some systems yes, through adaptation to the personal dictionary and the terms the user often uses; in others the system is static and does not adjust to a specific person without manual dictionary setup.

Why does voice input confuse words that sound similar?

Homophones (words that sound the same but are written differently) are a known limitation of any ASR: the model picks the most probable option from the context of the phrase and can be wrong without enough context; a personal dictionary and spelling out letters aloud reduce the frequency of such errors for specific terms.

How much faster is voice input than a keyboard?

Potok states about 900 characters per minute by voice against about 200 when typing, roughly 4 times faster. This is the product's own estimate, not an independent measurement.

Is there voice input on a computer without installing programs?

Yes: Windows has built-in voice input on Win+H (Russian only in Windows 11, there is no Russian in Windows 10; it does not work without the internet), and on a Mac there is system dictation in the keyboard settings. Both are free.

How to start

Potok is available for Windows, macOS and Linux. For two weeks everything is open for free, after that 2500 words per week without paying. Potok is a cloud service: speech is recognized on servers in Russia, and it does not work without the internet.

Download Potok

One subscription, two tools: dictation in any program on your computer and @potok_now_bot in Telegram, which transcribes voice messages, audio and video.

Related software: terms compared

Terms are taken from the services' official websites.
ServicePlatformsFreePricePayment from RussiaWhere speech is recognized
PotokWindows, macOS, LinuxFree: 2500 words per week; full access for the first 14 days, no card neededPro 690 ₽/mo, 499 ₽/mo when paid yearlyRussian bank cards, rublescloud, servers in Russia; internet required
SpeakFlowWindows, Web30 minutes of dictation per month, free forever690 ₽/mo (Personal (unlimited))rubles directlycloud (the data center country is not disclosed on the website) (not confirmed)
DiktuyWindows, macOS30 minutes of dictation per month, free forever239 ₽/mo (Profi (monthly equivalent when billed yearly)); 2870 ₽/yr (Profi (year, billed in full)); 479 ₽/mo (Max (monthly equivalent when billed yearly))rubles directly (not confirmed)cloud (the data center country is not disclosed on the website) (not confirmed)
TalkpadmacOS, Windows, Android2500 words/week on desktop, 1500 words/week on mobile, free forever; plus a 14-day Pro trial$6/mo (Pro (billed yearly, $72/yr)); $8/mo (Pro (monthly billing))unknown (not confirmed)cloud (the legal entity is registered in Australia; the website is hosted on Cloudflare; the exact location of the ASR servers is not disclosed in the privacy policy) (not confirmed)
VoistyWindows, macOS5 minutes of transcription (to judge the quality)$6.7/mo (Pro); $67/yr (Pro (year)); $99/one-time (Lifetime)Visa/Mastercard/Amex cards, PayPal; whether Russian cards are accepted is not stated (not confirmed)device (offline, local models such as Whisper Large V3 Turbo / NVIDIA Parakeet)
WhispeRuWindows7-day free trial (no card)290 ₽/mo (monthly license); 1990 ₽/yr (yearly license); 4990 ₽/one-time (lifetime license)rubles directly (payment via YooKassa)device (offline)
GigaTypeWindows, macOSfree with no limits: open source under the MIT license (Copyright (c) 2026 Disrupt Builders), all features are available without payment—not needed: there are no paid plansdevice (offline, Sber's GigaAM neural network)
MolviWindows, macOSfree forever, no accounts and no payment—not needed: the product is free; voluntary donations are accepted via Telegram Tribute or by card on the website, amounts are not fixeddevice (offline; best quality on an NVIDIA GPU, otherwise a fast model on the CPU)

Built-in dictation and Potok

Based on Microsoft and Apple help pages.
ToolWhere it worksRussian languageWhere speech is recognized
Windows voice typing (Win+H)Windows 10 and 11Windows 11 onlyMicrosoft cloud, does not work offline
macOS DictationmacOSyeson the device or on Apple servers
PotokWindows, macOS, Linuxprimary languagecloud, servers in Russia, does not work offline

Primary sources: speakflow.ru · speakflow.ru · speakflow.ru · blog.speakflow.ru

Mark Vesyoly — founder of Potok. Writes about Russian voice input: how to dictate in work apps, what gets in the way and how to fix it.
© Potok · Vladimir Olegovich Lipatnikov · INN 502915859701 · Public offer Comparisons · FAQ · About Free voice input · Voice input for AI chats · Data and Russian privacy law · Voice input on Mac · Bot plans · 14 days of Pro for a review