SpeechPad or Potok: what to choose for dictation in Russian

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

SpeechPad transcribes already recorded audio and video up to 15 minutes long for free and works in a browser and on a phone. Potok (potok.now) is a desktop program for Windows and macOS that inserts dictated text straight into the field of any app, with a free limit of 2500 words per week.

SpeechPad lives in a browser: you open a tab on speechpad.ru, dictate text into the notepad and copy it where you need. The same service can transcribe ready audio and video: up to 15 minutes on the free plan, without dictating again. Potok works differently: a hotkey starts recording right in the active field of Word, Telegram or a CRM, without an intermediate window and without copying. Both services put a clear focus on Russian, but the tasks they are chosen for diverge, from a one-off spoken note to dictating letters all working day.

SpeechPad or Potok: what SpeechPad is valued for

SpeechPad covers a scenario for which Potok is not meant: transcribing already recorded audio or video. The free plan handles such files up to 15 minutes: meeting notes, a voice message, a piece of a lecture turn into text without dictating again. Potok has no such feature at all: it inserts only what a person says into the microphone right now, not what is already recorded on a disk.

The second difference is mobile apps. SpeechPad, according to the service's official blog, installs on Android and iOS, so you can dictate a note from a phone in a traffic jam or on a walk, and transcription is available from the same mobile screen. Potok is desktop only, for Windows, macOS and Linux: the scenario “took out the phone and spoke” does not work here.

SpeechPad's list of languages is wider. Besides Russian and English, Armenian, Azerbaijani, Georgian, Latvian, Lithuanian, Kazakh, Uzbek, Ukrainian, Estonian, German, French, Spanish and Italian are stated. For a team where someone dictates not only in Russian this may be decisive. Potok is tuned specifically for Russian speech and mixed Russian-English text, not for this whole list. A broader review of similar services is in the comparison of voice input programs.

Payment and where the data is stored

Payment in both services is built for a Russian user. SpeechPad accepts Robokassa and bank cards directly on speechpad.ru; for access from abroad the service has a separate site, voicenotebook.com, with payment through PayPal, and the scheme is thought out even for the case when rubles do not go through. The legal entity in the payment details is the sole proprietor Alexey Vladimirovich Shechkov. Potok is also a sole proprietor, a Russian one, and payment is by a card of a Russian bank in rubles.

Data is harder to compare honestly. Where SpeechPad's speech recognition servers are physically located is not stated on the website and was not found in open sources. It cannot be claimed that the data centers are “its own” or “foreign”: this is not confirmed. For Potok it is stated explicitly: recognition runs on servers in Russia, which matters both for the 152-FZ personal data law and for companies that must not send voice data beyond the country's perimeter.

SpeechPad's extended mode removes the 15-minute limit on transcription and opens integration with the Windows and Linux OS. No official price list with this sum was found on the website. The figure appears only in user comments and reviews, so the exact cost should be checked on the website itself before paying, and not taken from figures in third-party sources. The cost of both services is in the table below.

How each handles Russian

For SpeechPad, Russian is the service's main language; the full list of the other fourteen supported languages is in the section above.

Potok is trained on its own model specifically for Russian speech, including names, professional terms and text where Russian is mixed with English (a frequent scenario in IT and marketing). On the internal GolosTest test Potok has a WER of 2.1%: this is our figure on our test; no independent comparisons of Russian recognition accuracy between Potok and SpeechPad have been published, so it cannot be said directly who is more accurate.

SpeechPad has a wider list of supported languages, while Potok has its own model for Russian speech with a personal dictionary for terms and names. The personal dictionary remembers the words you need once, and after that recognition puts them in without manual corrections at every dictation. This is especially noticeable where there are many specific terms and names.

Sources: speechpad.ru · speechpad.ru · speechpad.ru

When it is better to stay with SpeechPad

There are three cases where SpeechPad is the more logical choice. First: an already recorded file, a voice message from a messenger, audio from a meeting, a piece of video. SpeechPad will transcribe it outside live dictation, while Potok cannot do this at all. It inserts only what is said into the microphone right now.

The second case is a phone. If your working tool is Android or an iPhone, not a laptop, Potok has no apps for these platforms, only clients for Windows, macOS and Linux. The third is a language other than Russian: if the team dictates in Armenian, Kazakh or Georgian, SpeechPad's list of languages is wider. A similar choice between a notepad in a browser and a desktop client is covered in the comparison of Molvi and Potok.

It is worth remembering Potok's own architecture too: it is a cloud service, recognition runs on servers in Russia, and it does not work at all without the internet. For offline dictation you need a separate local tool. About SpeechPad there is nothing to say in this respect: where its recognition servers are located is not stated on the website.

Frequently asked questions

Primary sources: speechpad.ru · speechpad.ru · speechpad.ru