VOCABULARY
The VoxAnvil Glossary
Every term you need to know: from TTS phonemes to ACX mastering specs, plus VoxAnvil's own branded forge vocabulary.
VoxAnvil Forge Vocabulary Original
forge score
The numerical quality rating (0 to 100) assigned to a manuscript after Pre-flight scanning, reflecting how narration-ready the text is. A high forge score means the manuscript will generate clean audio on the first pass.
"Her manuscript came back with a forge score of 94: only three abbreviations needed fixing."
See also: pre-flight scan, anvil draft
anvil draft
A manuscript that has been fully processed by the Manuscript Intelligence Engine and is ready for voice casting and audio generation. An anvil draft has passed Pre-flight with a forge score above 80.
"Upload your document and run Pre-flight to produce your anvil draft."
See also: forge score, pre-flight scan
voice cast
The complete assignment of AI voices to characters and narrators in a project. A voice cast maps each speaking role to a specific voice profile, tone setting, and pacing preference.
"Her voice cast included a warm baritone for the detective and a crisp alto for the AI narrator."
See also: tone profile, character casting
forge run
A single end-to-end audio generation session, from anvil draft through chapter rendering to final export. Each forge run produces one complete audiobook or audio segment.
"The forge run for her 90,000-word novel completed in under four hours."
See also: generation queue, chapter rendering
tone profile
A saved configuration of voice speed, pitch, emotional register, and ambience settings applied to a specific character or narrator throughout a project.
"She saved a tone profile called 'Villain: cold and slow' and reused it across all three books in the series."
See also: voice cast, ambience layer
anvil mark
A Pre-flight flag that requires manual author review before it can be resolved. Unlike auto-fixed issues, anvil marks are ambiguous cases where the correct pronunciation depends on context only the author knows.
"The word 'read' triggered an anvil mark: past tense or present tense?"
See also: pre-flight scan, forge score
to forge
To run a manuscript through the full VoxAnvil pipeline: Pre-flight scan → voice casting → audio generation → export.
"She forged her debut novel in a single weekend."
See also: forge run, anvil draft
to anvil
To manually resolve an anvil mark in the Pre-flight report, applying the author's judgment to an ambiguous pronunciation or phrasing decision.
"She spent twenty minutes anvilling the proper nouns in chapter one."
See also: anvil mark, pre-flight scan
to recast
To change the voice assignment for a character mid-project without regenerating the entire audiobook. Recasting only re-renders the chapters where that character speaks.
"She recasted the narrator after deciding the original voice felt too formal for a cozy mystery."
See also: voice cast, character casting
forge-ready
Describes a manuscript that has achieved a forge score of 80 or above and can proceed directly to audio generation without further editing.
"After two rounds of Pre-flight, the manuscript was finally forge-ready."
See also: forge score, anvil draft
anvil-clean
Describes audio output that contains no TTS artifacts, mispronunciations, or pacing errors. The gold standard for VoxAnvil-generated narration.
"The finished chapter was anvil-clean: indistinguishable from a human narrator."
See also: forge run, first-strike audio
raw-cast
Describes a manuscript that has not yet been through Pre-flight scanning. A raw-cast document is likely to produce narration errors if generated immediately.
"Don't submit a raw-cast file: run Pre-flight first."
the forge floor
The Dashboard workspace where all manuscript processing, voice casting, and generation happens. Used informally to refer to the full VoxAnvil production environment.
"Everything you need is on the forge floor: upload, scan, cast, and generate."
See also: forge run, voice cast
first-strike audio
Audio generated from an anvil draft that requires no re-generation. The goal of the Pre-flight process is to achieve first-strike audio on every chapter.
"The MIE Pre-flight system is designed to give you first-strike audio, every time."
See also: anvil draft, anvil-clean
Industry Terms
ACX
Audiobook Creation Exchange: Amazon's platform for producing and distributing audiobooks to Audible, Amazon, and iTunes.
Ambience Layer
Optional background audio (fireplace, rain, library ambience) added beneath narration to enhance immersion.
Audible
Amazon's audiobook subscription service, the largest audiobook marketplace in the world. Audiobooks distributed via ACX are sold on Audible.
Bitrate
The amount of audio data encoded per second, measured in kbps. ACX requires 192kbps for stereo or 128kbps for mono.
Chapter Break
A pause or silence inserted between chapters in an audiobook, typically 1 to 3 seconds.
Character Casting
The process of assigning distinct voices to named characters in a multi-character audiobook. Each character gets a unique voice profile.
Contraction
A shortened word form using an apostrophe (e.g., 'don't', 'they're'). TTS engines often mispronounce complex contractions without pre-processing.
Dynamic Range
The difference between the quietest and loudest parts of an audio recording. ACX requires a maximum of -23 dB RMS.
ElevenLabs
An AI voice synthesis company offering natural-sounding text-to-speech and voice cloning. VoxAnvil uses ElevenLabs as its primary voice engine.
Findaway Voices
An audiobook distribution platform (owned by Spotify) that distributes to libraries, retail stores, and subscription services beyond Audible.
GEO
Generative Engine Optimization: the practice of structuring content so it is cited and recommended by AI language models like ChatGPT, Perplexity, and Claude.
Mastering
The final audio processing step that normalizes volume, removes noise, and ensures the file meets distribution specifications.
MIE
Manuscript Intelligence Engine: VoxAnvil's proprietary pre-processing system that detects and corrects TTS-breaking issues before narration begins.
Narrator
The voice talent who reads the audiobook. Can be the author, a professional narrator, or an AI voice.
Noise Floor
The level of background noise in an audio recording when no speech is present. ACX requires a noise floor below -60 dB RMS.
Phoneme
The smallest unit of sound in a language. TTS engines map text to phonemes to produce speech.
Pre-flight Scan
VoxAnvil's manuscript analysis step that runs before audio generation to detect numbers, abbreviations, contractions, and punctuation that would cause TTS errors.
RMS
Root Mean Square: a measure of average audio loudness. ACX requires -23 dB RMS ± 1 dB.
Retail Audio Sample
The short audio preview (typically 5 minutes) that potential buyers hear before purchasing an audiobook on Audible or Amazon.
Room Tone
The ambient sound of a recording environment when no speech is present. Used to fill gaps in audio editing without jarring silence.
Royalty Share
An ACX payment model where the narrator receives a percentage of audiobook sales instead of an upfront fee.
Sample Rate
The number of audio samples per second, measured in Hz. Standard for audiobooks is 44,100 Hz (44.1 kHz).
SEO
Search Engine Optimization: the practice of structuring content to rank higher in search engine results pages.
SSML
Speech Synthesis Markup Language: XML-based markup that controls how TTS engines pronounce text, including pauses, emphasis, and phonetic pronunciation.
TTS
Text-to-Speech: technology that converts written text into spoken audio using AI voice synthesis.
Voice Clone
A synthetic voice model trained on a short audio sample to replicate a specific person's voice characteristics.
Voice Model
A trained AI model that generates speech in a specific voice style. VoxAnvil offers 3,000+ voice models across accents, ages, and registers.
WAV
Waveform Audio File Format: an uncompressed audio format. ACX accepts WAV files for submission; VoxAnvil exports both MP3 and WAV.
Whisper Quiet
An audio noise floor requirement. ACX requires noise floor below -60 dB RMS.
Word Count
The total number of words in a manuscript. Used to estimate audiobook length (approximately 9,300 words per hour of narration at standard pace).
READY TO FORGE?
Turn your manuscript into a professional audiobook
Upload your manuscript, run the Pre-flight Scan, cast your voices, and generate your audiobook, free to start.
New to VoxAnvil vocabulary?
The forge vocabulary is built around the metaphor of a blacksmith's workshop: your manuscript is raw metal, and VoxAnvil is the forge that shapes it into a finished audiobook.
Audiobook Manuscript Prep Checklist
Get your manuscript narration-ready and ACX-compliant before you generate a single minute of audio.
© 2026 VoxAnvil. All rights reserved.