| Version | Year | Status |
|---|
| DAISY 2.02 | 2001 | Legacy, still used by older players |
| DAISY 3.0 (ANSI/NISO Z39.86-2005) | 2005 | Current standard |
| DAISY 3.1 | 2012 | Refinements |
| EPUB3 with DAISY profiles | 2014 | Convergence with EPUB3 |
For new productions: DAISY 3.0 or EPUB3 with DAISY accessibility profile. The two have largely converged; many "DAISY" books in 2026 are EPUB3 with DAISY-compliant authoring.
A DAISY book is a directory containing:
package.opf: manifest (similar to ePub's content.opf)ncc.html: navigation control center (table of contents)*.xml: text content files*.smil: synchronization between text and audio*.mp3 or *.wav: audio files*.png: optional images
The structure mirrors ePub3 with media overlays but with additional metadata for accessibility tools.
| Tool | Cost | Notes |
|---|
| Tobi (open source) | Free | Reference DAISY 3.0 authoring |
| EasyReader Plus | Paid | Library-grade tool |
| MyStudio PC | Paid | Commercial DAISY authoring |
| DAISY Pipeline 2 (open source) | Free | Conversion between formats |
| Calibre with DAISY plugin | Free | Limited DAISY support |
For most authoring: Tobi (free, multi-platform) or DAISY Pipeline 2. The Pipeline tool is also useful for conversions (EPUB to DAISY, Word to DAISY, etc.).
DAISY Pipeline 2 is the standard conversion tool:
# Install
git clone https://github.com/daisy/pipeline.git
cd pipeline
./build.sh
# Convert EPUB to DAISY 3.0
./bin/daisy-pipeline epub-to-daisy3 \
--input book.epub \
--output ./daisy-out/ \
--audio-encoder mp3
The Pipeline includes:
- EPUB to DAISY 3.0 conversion
- Word DOCX to DAISY conversion
- DAISY to EPUB conversion
- Validation against accessibility specs
- Audio synthesis from text (TTS) for books without recorded narration
For programmatic batch work, see Batch Processing Files Guide.
DAISY supports three "tiers" of synchronization detail:
- Phrase-level: each sentence or paragraph is highlighted with audio
- Word-level: each word highlights as audio plays
- Sentence + word: hybrid; sentence highlights, words underline
For most audiobook delivery: phrase-level. Word-level is labor-intensive (more SMIL data) but useful for educational materials and language learning.
DAISY accepts:
- MP3 at 32 kbps mono (smallest, for older hardware)
- MP3 at 64 kbps mono (production default)
- MP3 at 128 kbps stereo (high quality)
- AMR-WB (telephone-quality, very small)
- WAV (uncompressed, for older players)
For libraries serving low-bandwidth users: 32-64 kbps mono. For users with high-quality playback: 128 kbps stereo. Default to 64 kbps mono for general distribution.
For background on audio bitrate, see Audio Bitrate Quality Guide.
DAISY books come in two flavors:
Recorded narration: human reads the book, audio is captured. Quality is best, cost is high (1 hour audio per 9000 words typical).
TTS synthesis: audio generated from text by speech synthesis. Faster and cheaper. Modern neural TTS (Microsoft Azure Speech, Amazon Polly, Google WaveNet) approaches human quality.
For libraries serving visually impaired readers: recorded narration is preferred. For lower-cost catalogs: TTS is acceptable.
For TTS-based DAISY:
# DAISY Pipeline 2 with TTS
./bin/daisy-pipeline html-to-daisy3 \
--input chapter.html \
--output ./output/ \
--audio-encoder mp3 \
--tts-engine azure-speech \
--tts-voice en-US-AriaNeural
The Pipeline's TTS integration produces synchronized audio without manual recording.
Before delivery, validate with the Ace Accessibility Checker (free) and DAISY's reference validator:
# DAISY Pipeline validation
./bin/daisy-pipeline validate-daisy3 \
--input ./daisy-out/
# ACE accessibility check
ace --outdir ./ace-report/ ./daisy-out/
These check for:
- DAISY 3.0 spec compliance
- Audio file integrity
- SMIL synchronization accuracy
- Accessibility metadata completeness
- Navigation completeness
A passing report doesn't guarantee perfect user experience but catches structural errors that libraries reject.
Major distributors:
| Distributor | Region | Catalog requirements |
|---|
| Bookshare | US, global | Paid membership, qualified disability |
| RNIB Bookshare | UK | UK accessibility services |
| NLS BARD | US (Library of Congress) | US-only, certificate of disability |
| Vision Australia Library | Australia | Australian residents |
| Library of Lithuania for the Blind | Lithuania | Lithuanian residents |
Each distributor has its own submission requirements. Most accept DAISY 3.0 or EPUB3-with-DAISY-profile. For commercial publishers, the distribution flow goes through accessibility services rather than direct.
Audio out of sync with text: SMIL timestamps wrong. Re-synchronize with DAISY Pipeline's TTS-based alignment, or manually verify a few key transitions.
Validation fails on missing alt text: every image must have alt text. Add via authoring tool or programmatically before validation.
Chapter navigation broken: NCC (navigation control center) malformed. Re-generate with authoring tool from the source content.
File too large for distribution: audio bitrate too high. 32-64 kbps mono is typical. Higher bitrates only for high-quality variants.
Player hardware doesn't recognize the format: legacy DAISY 2.02 player. Author in 2.02 for old hardware compatibility, or use 3.0 for modern players.
For accessibility-focused PDF work, see Searchable PDF With OCR.
DAISY-specific players exist (Victor Reader, BookSense) and serve users who prefer dedicated audio hardware. EPUB3 covers most software-based reading. For library distribution: often DAISY required.
Yes via DAISY Pipeline 2. The conversion creates SMIL synchronization between the audio and a text version of the book. Quality depends on audio source.
For Bookshare and RNIB Bookshare: yes, with the synthetic origin disclosed. For NLS BARD: typically no, recorded narration only.
For a 50,000-word novel: 2-4 days for synchronization (with TTS), 6-8 days with manual review. With recorded narration: add the recording time (8-12 hours of studio).
iPhone: DAISY Reader (Apple Books doesn't natively read DAISY). Android: VoiceBooks, EasyReader. Most reader apps require purchase but are accessible-tax-free in many regions.
DAISY 3.0 books have multiple extensions. The package container can be .daisy3 or .epub (DAISY-EPUB hybrid) or unpacked as a directory. Players auto-detect.
For DAISY accessible audiobook delivery: author in Tobi or DAISY Pipeline 2, use 64 kbps mono MP3 for audio, validate with Ace and DAISY Pipeline's checker, distribute through Bookshare or RNIB Bookshare. EPUB3 with DAISY accessibility profile is the modern equivalent for many publishing workflows. Our audio converter handles audio bitrate adjustments before DAISY packaging.
