# Talking book glossary

This glossary explains the terms as used in this proposed production system. The linked articles and standards contain the implementation detail.

| Term | Meaning |
|---|---|
| NLS | The Library of Congress National Library Service for the Blind and Print Disabled. |
| DTB | Digital talking book: audio and supporting structure that allow accessible reading and navigation. |
| DAISY | Standards and technology for accessible publications, including structured talking books. |
| ANSI/NISO Z39.86 | The talking-book standard cited in the procurement; the edition matters. |
| `audioNCX` | A multimedia type describing full audio with navigation structure; it does not by itself promise a synchronized full-text product. |
| TTS | Text to speech: generating spoken audio from text. |
| ASR | Automatic speech recognition: generating a transcript from audio. |
| Forced alignment | Estimating where a supplied transcript fits the audio; distinct from independently recognizing what was spoken. |
| Canonical book model | The shared structured record linking original sources, approved text, navigation, audio and evidence. |
| Provenance | Records of where an item came from and how it was transformed. |
| Source span | An identifiable passage or audio range in an original source. |
| Spoken form | The approved text representation intended for narration, such as expanding a currency amount. |
| Lexicon | A book-level or shared set of pronunciation decisions. |
| Inference | Running a trained model to generate or analyze content. |
| Model weights | Numerical parameters of a trained model; they are distinct from application code and voice assets. |
| Deterministic compiler | Conventional software that produces prescribed files from explicit accepted inputs and a versioned rule set. |
| Output profile | The exact standards, media, metadata and packaging rules chosen for a build. |
| OPF | The package document describing metadata, included resources and organization. |
| NCX | Navigation Control file for XML; defines navigation structures and labels. |
| SMIL | Synchronized Multimedia Integration Language; links playback structures to media segments. |
| DTD | Document Type Definition; a schema mechanism used by these XML formats. |
| Entity | An XML declaration/reference whose allowed definitions must be controlled and packaged as required. |
| navPoint | A hierarchical navigation point, such as a chapter. |
| navTarget | A target in an auxiliary navigation list; its allowed use depends on the standard edition. |
| Segment | A useful navigable unit that need not be a chapter or heading. |
| PCM / WAV | PCM represents uncompressed audio samples; WAV is a common container for those samples. |
| Sample position | An integer location in audio at a known sample rate, useful for precise internal timing. |
| AMR-WB+ | The extended wideband audio codec required by the cited NLS profile; different from AMR-WB. |
| 3GP | The container used for the compressed audio, with NLS-specific constraints. |
| LUFS | A unit for measured perceived loudness; the applicable mastering edition defines the target. |
| PDTB | Protected Digital Talking Book; content protection and authorization for the book/player ecosystem. |
| MD5 | A checksum algorithm used in the specified DTB checksum format. A checksum is not an authorization or authenticity system. |
| CRC | A ZIP integrity check value; the delivery specification constrains how it appears in the headers. |
| Idempotent job | A job whose retry does not create unwanted duplicate effects. |
| Incremental build | Regenerate changed outputs and their dependencies while reusing unaffected accepted artifacts. |
| S3 | AWS object storage; proposed for internal source and artifact storage here. |
| GPU | A processor commonly used to accelerate model inference. |
| Real-time factor | Processing time divided by audio duration under defined hardware and workload conditions. |
| POP | Period of performance: the contractual time window for performing the work. |
| SOW | Statement of work: the document defining requested work and deliverables. |
| RFI | Request for information: a market-research request, distinct from an award or accepted implementation. |

For precise format definitions, use [J2](https://sam.gov/api/prod/opps/v3/opportunities/resources/files/f14f0acd77f64c29ad4d40567abbfab1/download) or [the 2002 standard](https://www.daisy.org/z3986/specifications/Z39-86-2002.html) according to the approved profile. General operational definitions in this table are explanatory additions.