Anki Enhancement

Anki Enhancement is GSM's core feature. It monitors Anki for newly created cards and automatically enhances them with audio, screenshots, and more — turning a basic vocabulary card into a rich, context-filled flashcard.

A card before and after GSM enhancement — audio, screenshot, and game context added automatically

What Can Be Added

When you mine a word in Yomitan and a new card appears in Anki, GSM can automatically add:

Media	Description
Sentence Audio	Audio trimmed from the game recording using Voice Activity Detection (VAD).
Screenshot	A static screenshot from the game at the moment the sentence appeared.
Animated Screenshot	A short animated clip (AVIF/WebP) instead of a static image.
Previous Sentence Screenshot	Screenshot from the prior line, for additional context.
Video Clip	A short video clip with audio of the relevant game moment.
Sentence with Furigana	Auto-generated ruby text for the sentence.
Game Name	The current game, added as a tag or field value.
AI Translation	A context-aware translation via your configured AI provider.
Tags	Custom tags, game tags, and NSFW tags.

How It Works

1. Card Detection — GSM polls Anki-Connect at a configurable rate (default: every second) to detect newly added cards.
2. Line Matching — The new card's sentence is matched against GSM's text log to find the corresponding game line and its timestamp.
3. Replay Buffer — GSM triggers OBS to save the replay buffer, capturing the relevant game footage.
4. Media Generation — Audio is extracted via FFmpeg and trimmed with VAD. Screenshots are captured at the sentence's timestamp.
5. Confirmation Dialog — If enabled, a dialog appears where you can review and edit before the card is updated.
6. Card Update — Media files are uploaded to Anki's collection and the card fields are updated via Anki-Connect.

tip

If you mine a second word from the same sentence, GSM reuses the media from the first card instead of regenerating it.

Confirmation Dialog

The confirmation dialog lets you review and edit before the card is updated

When Show Update Confirmation Dialog is enabled, GSM shows a dialog before updating each card. You can:

• Edit the sentence text and translation.
• Retake or adjust the screenshot.
• Play or replace the audio.
• Toggle the NSFW tag.
• The dialog auto-accepts after a configurable timer (default: 10 seconds).

Texthooker Integration

The texthooker page provides several tools that work alongside Anki Enhancement:

Texthooker buttons for multi-line mining, screenshots, audio playback, and AI translation

• Checkboxes — Combine multiple lines when mining. The resulting card will have text from each selected line, as well as the full sentence audio for all lines.
• 📷 Screenshot — Opens a frame selector to hand-pick the screenshot from frames around the voiceline.
• 🔈 Audio — Plays back the voiceline audio. Subsequent presses are much faster (cached).
• 🌐 Translate — When AI is configured, places a context-aware translation underneath the line.

Key Settings

Anki Connection

Anki connection and field mapping configuration

Setting	Description	Default
Enabled	Enable Anki integration	true
Update Anki	Enable card enhancement	true
Anki-Connect URL	URL of the Anki-Connect server	http://127.0.0.1:8765
Note Type	The Anki note type to monitor	—
Polling Rate	How often to check for new cards (ms)	1000

Field Configuration

Each Anki field (sentence, audio, picture, etc.) has individual controls:

Option	Description
Enabled	Whether GSM should populate this field.
Overwrite	Whether to replace existing content in this field.
Append	Whether to add to existing content instead of replacing.

The field names must match your Anki note type exactly (case-sensitive).

Confirmation

Setting	Description	Default
Show Confirmation Dialog	Show a review dialog before updating	—
Auto-Accept Timer	Seconds before auto-accepting the update	10

Tags

Setting	Description	Default
Custom Tags	Tags added to every card	GSM
Add Game Tag	Auto-add the current game name as a tag	true
Parent Tag	Prefix for game tags (e.g., Game::GameName)	Game
Tags to Work On	Only update cards that already have these tags (useful for filtering)	—
Tag Unvoiced Cards	Tag cards where VAD found no voice	—

Audio Fine-Tuning

Audio offset and VAD settings for fine-tuning trimmed audio

Setting	Description	Default
Beginning Offset	Time offset for the start of audio extraction (usually negative)	-0.5s
End Offset	Time added after the VAD trim result	0.0s
VAD Trim Beginning	Let VAD trim the start of the audio	false
External Audio Tool	Path to an external editor (e.g., OcenAudio) for manual trim	—

Screenshot Timing

Setting	Description	Default
Screenshot Timing	When to capture relative to the line: beginning, middle, or end	beginning
Screenshot Offset	Seconds to offset from the timing anchor	1.0
Use Screenshot Selector	Show a frame picker for every card	false

Related Settings

These settings in other tabs affect how media is generated for card enhancement:

• Screenshot tab: Format, quality, resolution, animated screenshots.
• Audio tab: Format (mp3, opus, ogg), quality, FFmpeg presets.
• VAD tab: Which VAD model to use (Silero or Whisper) and sensitivity.
• AI tab: Translation provider and prompt settings.
• Features tab: Full Auto mode, notifications, open card in browser.

Troubleshooting

Audio is cut off at the beginning

This is usually due to a poorly timed text hook event. Increase the negative Beginning Offset in the Audio tab. You can also enable VAD Trim Beginning if you need a large offset but there's silence at the start.

Sentence field doesn't match the full line

This is a Yomitan behavior, not GSM. Yomitan uses sentence termination characters (like 。) to determine where a sentence begins and ends. Set up a Yomitan profile for the texthooker page with Sentence Termination Characters set to Newlines only.

GSM is updating cards from other sources

Use the Tags to Work On setting to restrict GSM to only cards with a specific tag. Combine this with Yomitan profiles to tag cards differently per source.

Prerequisites

• Anki running with the Anki-Connect add-on installed (default port 8765).
• OBS connected with replay buffer enabled and set to 60–120 seconds.
• A dictionary workflow (e.g., Yomitan) that creates cards with a word/sentence field.