Files

Aodhan Collins 66749a5ce7 Initial commit

2025-10-06 00:33:04 +01:00

6.4 KiB

Raw Blame History

TTS Quality Controls

Overview

EVE now includes comprehensive voice quality controls allowing you to customize speed, stability, and clarity of text-to-speech output.

Controls Available

1. Speed Control (All Voices)

Range: 0.25x - 4.0x
Default: 1.0x (Normal)
Applies to: Both Browser TTS and ElevenLabs

0.25x - 0.75x: Slower speech, good for learning or understanding complex content
1.0x: Natural speaking pace
1.25x - 2.0x: Faster speech, efficient for experienced listeners
2.0x - 4.0x: Very fast, for quickly scanning content

2. Stability Control (ElevenLabs Only)

Range: 0% - 100%
Default: 50%
Applies to: ElevenLabs voices only

What it does:

Controls consistency vs expressiveness of the voice
Higher values = more consistent, predictable delivery
Lower values = more varied, emotional, expressive

When to adjust:

High (70-100%): Audiobooks, technical content, professional narration
Medium (40-60%): General conversation, balanced approach
Low (0-30%): Character voices, dramatic readings, creative content

3. Clarity Control (ElevenLabs Only)

Range: 0% - 100%
Default: 75%
Applies to: ElevenLabs voices only

What it does:

Controls similarity boost / voice clarity enhancement
Higher values = closer to original voice, enhanced clarity
Lower values = more variation, creative interpretation

When to adjust:

High (70-100%): Maximum clarity, important information, professional use
Medium (50-70%): Natural balance
Low (0-40%): More creative interpretation, character variation

User Interface

Location

Settings > Voice Settings > Voice Quality Settings

Design

Speed: Full-width slider with 0.25 step increments
- Shows current value in label (e.g., "Speed: 1.50x")
- Visual markers at 0.25x, 1.0x, and 4.0x
Stability: Full-width slider with 5% step increments
- Shows percentage in label (e.g., "Stability: 50%")
- Disabled (grayed out) when using browser voices
- Helpful description below slider
Clarity: Full-width slider with 5% step increments
- Shows percentage in label (e.g., "Clarity: 75%")
- Disabled (grayed out) when using browser voices
- Helpful description below slider

Smart UI Features

ElevenLabs-only controls show "(ElevenLabs only)" in label
Controls are disabled when browser voice is selected
Real-time value display as you drag sliders
Settings persist across sessions
All controls visible even when disabled for easy reference

Technical Implementation

Settings Store

ttsSpeed: number // 0.25 to 4.0
ttsStability: number // 0.0 to 1.0
ttsSimilarityBoost: number // 0.0 to 1.0

Usage in TTS

await ttsManager.speak(text, {
  voiceId: selectedVoice,
  volume: 1.0,
  rate: ttsSpeed,         // Browser TTS rate
  stability: ttsStability, // ElevenLabs stability
  similarityBoost: ttsSimilarityBoost // ElevenLabs clarity
})

Provider-Specific Application

Browser TTS:

Uses rate parameter from speed control
Ignores stability and similarity boost (not applicable)

ElevenLabs TTS:

Applies all three parameters
Speed can be adjusted post-processing if needed
Stability and similarity boost sent directly to API

Examples

For Audiobooks

Speed: 1.0x - 1.25x (comfortable listening)
Stability: 80% (consistent narration)
Clarity: 85% (clear pronunciation)

For Casual Chat

Speed: 1.0x (natural pace)
Stability: 50% (balanced)
Clarity: 75% (good clarity)

For Quick Scanning

Speed: 2.0x - 3.0x (fast playback)
Stability: 60% (maintain clarity at speed)
Clarity: 90% (maximum clarity for comprehension)

For Character Voices

Speed: 0.75x - 1.0x (theatrical pacing)
Stability: 20% (high expressiveness)
Clarity: 50% (allow variation)

Benefits

✅ Personalization - Adjust voice to your preferences
✅ Accessibility - Slower speeds for comprehension
✅ Efficiency - Faster speeds for quick consumption
✅ Quality Control - Fine-tune ElevenLabs voice output
✅ Flexibility - Different settings for different use cases
✅ Universal - Speed works on all voices, premium controls for ElevenLabs

Persistence

All settings are:

✅ Saved to localStorage
✅ Persist across app restarts
✅ Applied automatically to all future TTS playback
✅ Can be changed at any time

Future Enhancements

Potential Additions

Pitch control for browser TTS
Volume control per-voice
Per-voice presets (save favorite settings for each voice)
Quick presets (Audiobook, Podcast, Speed Reader, etc.)
Real-time adjustment while audio is playing
A/B comparison to test settings side-by-side

Advanced Features

Voice EQ for fine-tuning frequency response
Emotion control for ElevenLabs (happy, sad, excited, etc.)
Speaking style selection (narration, conversation, etc.)
Prosody controls (emphasis, pauses, intonation)

Troubleshooting

Sliders Not Responsive

Check that voice is enabled
Verify a voice is selected
Try refreshing the settings panel

ElevenLabs Controls Disabled

Make sure an ElevenLabs voice is selected (starts with "elevenlabs:")
Browser voices won't enable these controls (by design)
Check that ElevenLabs API key is configured

Settings Not Saving

Check browser localStorage permissions
Try clearing cache and reloading
Verify settings store is persisting

Speed Not Applying

Browser TTS: Rate should change immediately
ElevenLabs: Speed adjustment may vary by voice
Try values between 0.5x - 2.0x for best results

Testing

To Test Speed Control

Enable TTS
Adjust speed slider
Click speaker icon on a message
Voice should speak at selected speed

To Test ElevenLabs Controls

Select an ElevenLabs voice
Adjust stability slider
Adjust clarity slider
Click speaker icon
Notice difference in voice quality

To Test Persistence

Adjust all sliders
Close settings
Restart app
Open settings
Values should be preserved

Recommended Settings

Default (Balanced):

Speed: 1.0x
Stability: 50%
Clarity: 75%

Professional:

Speed: 1.0x
Stability: 80%
Clarity: 85%

Expressive:

Speed: 1.0x
Stability: 30%
Clarity: 60%

Fast Listener:

Speed: 1.75x
Stability: 65%
Clarity: 90%

Status: ✅ Complete
Version: v0.2.0-rc
Date: October 5, 2025

6.4 KiB Raw Blame History