creator/cc-1c-skills
1
0
Fork 0
mirror of https://github.com/Nikolay-Shirokov/cc-1c-skills.git synced 2026-06-10 16:14:54 +03:00
Code Issues Packages Projects Releases Wiki Activity

fix(web-test): reduce smart TTS wait from 100ms/char to 70ms/char

Browse Source 
Measured real Edge TTS (ru-RU-DmitryNeural) durations:
  57 chars → 4.4s (77ms/char)
  72 chars → 6.0s (84ms/char)
  126 chars → 8.2s (65ms/char)
  745 chars → 48.0s (64ms/char)

Old 100ms/char overestimated by 30-55%, causing long silent pauses
after speech in showImage/showTitleSlide/showCaption. New 70ms/char
gives ~10% safety margin without excessive silence.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Nick Shirokov
2026-03-18 21:05:13 +03:00
parent dc2056a5d5
commit b68f4145ce
2 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.claude/skills/web-test/recording.md
+1 -1
View File
@@ -92,7 +92,7 @@ Display a text overlay on the page (visible in recording). Calling again updates
The overlay uses `pointer-events: none` — does not interfere with clicking.
**Smart TTS wait** (during recording): `showCaption` automatically pauses for the estimated TTS speech duration (~100ms per character, min 2s). The next `wait()` call accounts for this — if the explicit pause is shorter than the TTS wait already done, no extra delay is added. If longer, only the remaining difference is waited. This means script authors don't need to calculate TTS timing manually.
**Smart TTS wait** (during recording): `showCaption` automatically pauses for the estimated TTS speech duration (~70ms per character, min 2s). The next `wait()` call accounts for this — if the explicit pause is shorter than the TTS wait already done, no extra delay is added. If longer, only the remaining difference is waited. This means script authors don't need to calculate TTS timing manually.
### `hideCaption()`
.claude/skills/web-test/scripts/browser.mjs
+3 -3
View File
@@ -3939,7 +3939,7 @@ export async function showCaption(text, opts = {}) {
// Use video timeline position (accounts for frame duplication) instead of wall-clock
recorder.captions.push({ text, speech, time: Math.round(recorder.videoTimeMs) });
// Estimate TTS duration and wait so the video has enough screen time for voiceover
smartWaitMs = Math.max(2000, speech.length * 100);
smartWaitMs = Math.max(2000, speech.length * 70);
}
const position = opts.position || 'bottom';
const fontSize = opts.fontSize || 24;
@@ -4207,7 +4207,7 @@ export async function showTitleSlide(text, opts = {}) {
const captionText = typeof speech === 'string' ? speech : text.replace(/\n/g, ' ');
if (captionText) {
recorder.captions.push({ text: captionText, speech: captionText, time: Math.round(recorder.videoTimeMs) });
smartWaitMs = Math.max(2000, captionText.length * 100);
smartWaitMs = Math.max(2000, captionText.length * 70);
}
}
@@ -4315,7 +4315,7 @@ export async function showImage(imagePath, opts = {}) {
const captionText = typeof speech === 'string' ? speech : '';
if (captionText) {
recorder.captions.push({ text: captionText, speech: captionText, time: Math.round(recorder.videoTimeMs) });
smartWaitMs = Math.max(2000, captionText.length * 100);
smartWaitMs = Math.max(2000, captionText.length * 70);
}
}