Voice typing in Russian: built-in tools and separate programs

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

Voice typing in Russian works through the built-in system panel or through a separate app. In Windows this is Win+H, but only Windows 11 understands Russian there; in macOS it is the system Dictation with Russian support. Potok (potok.now) is a separate app with the Ctrl+Space hotkey and 2500 words per week on the free plan.

Voice typing is entering text by voice instead of a keyboard, and in Russian it does not work equally well everywhere. The built-in tools of operating systems support Russian differently: in some it is fully there, in others it is absent altogether, and separate programs add what the system lacks: a personal dictionary, one hotkey for all apps, a more accurate model for a specific language. The difference between these options determines which way is worth using for you.

What the system offers without installation

In Windows, voice typing in Russian is available through the panel opened with Win+H, with an important caveat: it understands Russian only in Windows 11. In Windows 10 Russian is not among the supported languages at all; according to Microsoft's official data, English, German, French, Spanish, Italian, Brazilian Portuguese and Chinese are available there. For a Windows 10 user who needs voice input specifically in Russian, the built-in panel will not solve the task in principle, and a separate program is needed right away.

In Windows 11 Win+H works differently: you turn on the key combination, select Russian in the speech recognition settings, and the panel is ready to work in any text field. Recognition there is in the cloud: the audio goes to a Microsoft server and is processed there, so without the internet the panel will not open, even though it is built into the system.

In macOS the situation is simpler: the system Dictation supports Russian regardless of the macOS version. It is turned on in System Settings, in the keyboard section, and also works in any text field without installing extra programs. Like the Windows panel, Dictation recognizes speech through a server, not on the computer itself, so it is unavailable without a network too.

What both built-in solutions have in common: they are free, need no installation and cover the basic scenario of entering text by voice. What they lack: a personal dictionary for terms and names, a separate hotkey for a specific task, and stable work on long technical texts, where a model tuned specifically for such speech is needed.

For the user this means a practical choice: if the task is to quickly dictate a short message once a day, a built-in tool is usually enough. If dictation becomes part of the daily work routine with professional vocabulary, the limitations of the built-in panels show quickly.

Separate programs and how they differ

Separate programs for voice typing in Russian differ from the built-in tools in several parameters, and the choice depends on which of them matter to you personally. None of these parameters matters equally to everyone: what is critical for a lawyer with hundreds of terms a day may not concern at all someone who dictates a couple of messages to a messenger.

The first difference is how much Russian is the main language for the recognition model, rather than one of thirty supported languages. GigaType, for example, is based on the GigaAM neural network from Sber, trained specifically for Russian. Talkpad is the opposite: it states multilingual auto-detect and more than a hundred languages with no separate model specifically for Russian. For mixed Russian and English speech and professional terms, this difference is felt faster than it seems at first glance.

The second: a personal dictionary. The ability to tell the program rare names, company names and professional terms in advance noticeably reduces recognition errors over time; the built-in tools usually have no such feature at all.

The third: one hotkey for all apps. A separate program is built in at the system level and inserts text into any active window, whether email, a messenger or a document, with the same key combination, without switching between different tools for different tasks. For Potok this is the Ctrl+Space hotkey.

The fourth: the payment model and free access. Some programs, such as SpeakFlow and Diktuy, keep a small free limit in minutes every month and then move to a subscription. Others, such as WhispeRu, give a short trial period with no card, after which access becomes fully paid. For Potok it is 2500 words per week with no card, and for the first 14 days the access is full and free.

The fifth: where recognition physically happens. For most of the listed programs it is the cloud, but not for all: GigaType and Molvi, for example, process speech right on the device, without sending audio to a server at all. The difference shows not in quality itself but in whether the internet is needed and where the audio goes; this is a separate selection criterion, not directly related to recognition accuracy.

The sixth: the accuracy stated by the developer. Some services publish their own figures: Diktuy, for example, states 95–98% accuracy on clean Russian speech in its own tests. This is the developer's own figure, not the result of an independent comparison, and it is better as a guide than for direct comparison between programs.

The seventh: features beyond dictation itself. SpeakFlow, for example, has a separate feature: dictation in Russian with insertion of a translation into English, something that neither the built-in solutions nor most of the listed programs have. In Lispr some features (push-to-talk, translation, dictionary, local history) always stay free, regardless of the limit of dictation minutes per week.

Where Potok is not the best option

Potok is a cloud service: recognition runs on servers in Russia, and it does not work at all without the internet, like the built-in tools of both systems. Those who are set on finding an offline solution need a separate local program, not a cloud service of any kind.

Potok does not support voice commands for editing text: it inserts what you said and does not edit what was written earlier. Transcription of already recorded files (meetings, calls, voice messages) is also not among its tasks: for that you need a separate transcription tool.

The service has no mobile apps for Android and iPhone, only Windows and macOS. The free limit is 2500 words per week: for those who dictate all working day it runs out quickly, and without moving to a paid plan, counting on free voice typing stops working in practice. More on how such programs differ from each other is in the article on a program for dictating text on a computer and in the question of whether a computer has voice input without installing programs.

For companies that forbid sending voice outside their own infrastructure, the ordinary cloud plan is not enough either: they need a variant with servers inside the company perimeter, available on separate request.

Frequently asked questions

Built-in dictation and Potok

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