Skip to content

First version

Choose a tag to compare

View all tags
@Pashtet495 Pashtet495 released this 20 Jun 14:51
rel
60851ed
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

EN

  • Supported operating systems: Windows 10/11.

  • Supported recognition languages: Bulgarian (bg), Croatian (hr), Czech (cs), Danish (da), Dutch (nl), English (en), Estonian (et), Finnish (fi), French (fr), German (de), Greek (el), Hungarian (hu), Italian (it), Latvian (lv), Lithuanian (lt), Maltese (mt), Polish (pl), Portuguese (pt), Romanian (ro), Slovak (sk), Slovenian (sl), Spanish (es), Swedish (sv), Russian (ru), Ukrainian (uk).

  • Supported interface languages: En, Ru, De, Fr, Es, It, Ch.

Minimum system requirements:

  • SSD/HDD: 1.5 GB

  • Without a graphics card or with a graphics card that doesn't support the Vulkan API:

CPU 4 cores 3.5 GHz
RAM: 6 GB

  • With an integrated graphics card or with a discrete graphics card that supports the Vulkan API:

CPU 2 cores 2 GHz
RAM: 6 GB
VRAM: 1 GB

When working with an HDD, there may be a 2-3 second delay between pressing the record button and recording starting.

Main screen:
001

Settings screen
002

  • Recommended "BACKEND" setting: GPU Vulkan
  • If the application runs in parallel with neural networks or other workloads that consume all video memory, it is recommended to select a value (0, 1, etc.) in the "GPU ID" field that uses integrated graphics (if integrated graphics are available). This will ensure that the voice recognition model is loaded into RAM rather than VRAM (Intel UHD Graphics 770 provides approximately 15x real-time performance for the duration of the audio being loaded).

Context menu in the tray.
003
When you click on the cross in the main interface, the program is minimized to the tray.

The program has two recognition modes:

  • with minimal delay
  • with better correction.
    In correction mode, the text is accumulated and processed by the neural network until the sentence is completed. The model removes repetitions, hesitations, etc.

RU

  • Рекомендуемая настройка Бэкенда - GPU Vuklan.
  • Если распознавание запускается параллельно с другими задачами использующими VRAM полностью - рекомендуется запускать на встроенной графике, в таком случае модель загружается в оперативную память (выбор устройства производится в настройках в поле ID GPU).
  • Программа при нажатии на крестик сворачивается в трей.
  • В программе два режима распознавания с микрофона:
    С минимальной задержкой
    Лучшее распознавание.
    В режиме лучшего распознавания модель накапливает фразу и исправляет её, убирая запинки, повторы, слова паразиты и т.д.
Assets 4
Loading