Choose a workspace
Open the dashboard and select the tool that matches the required output.
GiliSoft AI Toolkit User Guide
AI Toolkit User Guide for Windows
Choose the right AI workspace, add source material, review the result, and save accepted output.
- Extract and transform text and speech
- Create authorized voice and face projects
- Clean, cut out, repair, and enhance images
Complete an AI task in four steps
Start with one representative file or short sample. Check the result before processing important or larger source material.
Add the source
Load text, audio, an image, or another supported source and confirm it is correct.
Set the options
Choose language, voice, selection, quality, or processing settings for the task.
Review and save
Inspect the generated result, correct visible problems, and export it to a new file.
Choose the AI Toolkit workspace
Each workspace has a separate input and output. Use the task name below to open the correct tool instead of sending every source through the same process.
OCR
Extract editable text from screenshots, scans, photographed pages, and image-based documents.
Text to Speech
Turn written scripts into spoken audio for drafts, lessons, accessibility, and narration.
Speech to Text
Transcribe supported recordings into editable text for notes, interviews, and documentation.
Voice Clone
Create a custom voice only from recordings you own or have permission to use.
Face Swap
Prepare authorized photo and video face projects and inspect identity and edge consistency.
Background Cutout
Separate a subject from its background for product images, profiles, and design assets.
Image Cleanup
Remove unwanted marks or objects from images you are authorized to edit.
More AI Tools
Return to the dashboard for photo repair, enhancement, subtitle, and other available utilities.
Create text and voice output
Match the source, language, and output
- Open OCR, Text to Speech, Speech to Text, or Voice Clone from the dashboard.
- Add a clear source image, script, or audio sample.
- Select the correct language and available voice or recognition options.
- Process a short sample and check names, numbers, punctuation, pronunciation, and timing.
- Edit the source or settings when the sample needs correction.
- Export the approved text or audio to a clearly named file.
AI output can contain recognition or pronunciation errors. Review important names, dates, figures, and quoted material before use.
Prepare face and image results
Use clear source material and inspect the boundaries
- Open Face Swap, Background Cutout, Image Cleanup, Repair, or Enhancement.
- Load an image or video you own or are authorized to edit.
- Choose a source with clear subjects, useful resolution, and suitable lighting.
- Mark the required subject or cleanup area where the tool provides a selection control.
- Generate a preview and inspect hair, hands, face edges, text, reflections, and repeated textures.
- Save the result as a new file so the original remains available.
Do not use face or voice tools to impersonate another person or misrepresent consent. Keep the source and generated output identifiable in sensitive workflows.
Review AI output before delivery
Accuracy
Check text, speech, faces, object edges, colors, and important visual details.
Permission
Confirm you may use the source voice, face, image, audio, or document.
Originals
Keep the unmodified source until the generated result has been accepted.
Destination
Open the exported file in the application or device where it will be used.
AI Toolkit troubleshooting and FAQ
Why is OCR text inaccurate?
Use a sharper, straighter image with readable contrast, select the correct language, and review complex layouts manually.
Why does generated speech pronounce a word incorrectly?
Check the selected language and voice, adjust spelling or punctuation, and test a shorter sentence before exporting the full script.
Why does a cutout have rough edges?
Start from a larger image with clearer subject separation and inspect hair, transparent areas, and shadows at full size.
Why does a face result look unnatural?
Use sources with compatible angle, lighting, expression, and resolution, then inspect several frames when processing video.
Should I overwrite the original file?
No. Save generated output separately until it has been reviewed and accepted.
Where can I see every available tool?
Open the AI Toolkit dashboard or visit the GiliSoft AI Toolkit product page.
Ready to try an AI workflow?
Choose the matching workspace, add a representative source, review the generated result, and save the accepted output separately.
