You can love someone and still not know how to be in the same room with them. It starts small: when one person reaches with questions that feel like an interrogation, and the other retreats just to catch their breath. Two people who never ran out of love—only the vocabulary to touch without bruising each other.
CASE STUDY & AUDIO COMPANION//
Solving TTS Ear Fatigue in Long-Form Audiobooks
In producing the audiobook for The Chipped Mug, default voices were rejected from day one due to known robotic timbres. But building a custom generative voice revealed an intricate multi-stage challenge: from excessive breathing noise, to over-corrected studio flatness, to restoring dynamic range and achieving ACX-compliant mastering.
PHASE 01
The Breathy Custom Voice vs. Studio Isolation
Our initial custom voice had human warmth, but the generative model heavily over-indexed on respiratory artifacts. Below is the transition from noisy breath inhalations to clean studio isolation.
STAGE 1ACustom Voice Inception
4m 23s WAVRaw Custom Voice (Heavy Breaths & Artifacts)
0:004m 23s WAV
What to Listen For
Notice the prominent breath inhalations between phrases and trailing mouth clicks. While organic at first, these repetitive acoustic artifacts rapidly cause listener exhaustion over a 15-minute chapter.
Engineering Context
We refused to use default voices, but our first custom voice profile allowed too many micro-breaths and saliva artifacts into the vocal tract. Listen to the physical effort of each breath.
STAGE 1BTag & Parameter Tuning
4m 38s WAVCleaned Studio Vocal (Breaths Stripped)
0:004m 38s WAV
What to Listen For
Notice how the heavy breathing, mouth noises, and respiratory clicks have been completely eliminated. The voice is now crisp, clean, and isolated like a professional isolation booth.
Engineering Context
By adjusting speech-generation parameters and directive tags, we stripped away the excessive breath artifacts. But as you will hear in Phase 2, this introduced an unexpected new problem: digital flatness.
PHASE 02
Combating Studio Flatness (Breaking the Monotony)
Cleaning up the breaths solved the noise, but stripping the breathing intervals created an unnatural, sterile flatness. Without dynamic contrast, the brain quickly fatigued from acoustic uniformity.
STAGE 02Cadence & Range Injection
2m 23s WAVThe Peaks and Valleys Studio Test
0:002m 23s WAV
What to Listen For
Notice the deliberate pacing changes: phrases that accelerate, followed by sudden deceleration and breath pauses. We actively forced peaks and valleys back into the delivery to shatter the statistical metronome.
Engineering Context
Complete digital silence between words sounds dead to the human ear. In this test, we forced dynamic peaks and valleys into the prose delivery to re-establish narrative rhythm and contrast.
PHASE 03
The Final Persona: Garbor Expressive British
Combining the lessons of Phases 1 and 2: maintaining studio-clean isolation while baking natural dramatic gravel, chest resonance, and authentic dialect modulation directly into the voice profile.
STAGE 03Persona Engineering
2m 10s WAVGarbor Expressive British Studio Voice
0:002m 10s WAV
What to Listen For
Listen for subtle dramatic gravel, chest resonance, and conversational British inflection. It avoids Phase 1’s heavy breathing without succumbing to Phase 2’s over-calculated rigidity.
Engineering Context
This custom persona became our production anchor: clean enough for commercial distribution, yet nuanced enough to sustain hours of intimate relationship dialogue.
STAGE 04
The Chipped Mug: Full Commercial Audiobook
The production finish line: scene-level acting prompts + temperature 1.15 for prosodic entropy + 1.05x speed boost + 4-stage ACX-compliant DSP mastering chain.
STAGE 0413-Track Commercial Release
Now Streaming GloballyThe Chipped Mug: Full Commercial Audiobook
With temperature set to 1.15, the speech model generates subtle phoneme duration and pitch variations that mirror human spontaneity. Combined with our automated ACX mastering chain (80Hz highpass, soft expander gate, and two-pass EBU R128 normalization to -20.0 LUFS), the entire 13-track novel is available across all major streaming services.
Stream the full 13-track commercial audiobook:
DSP
The ACX Mastering Chain Benchmarks
Commercial audiobook platforms enforce strict acoustic thresholds. Here is how our automated audio pipeline achieved complete compliance:
| Acoustic Metric | Raw Model Output | Mastered Delivery | ACX / Audible Standard |
|---|---|---|---|
| Integrated Loudness | -24.8 LUFS (Too quiet) | -20.0 LUFS | -23.0 to -18.0 LUFS |
| True Peak | -0.8 dBTP (Clipping risk) | -6.5 dBTP | ≤ -3.0 dBTP |
| Noise Floor | -52.0 dB (Fails spec) | -68.4 dB | ≤ -60.0 dB RMS |
| Pacing / Flow | Sluggish, uniform drift | Dynamic (1.05x boost) | Natural human pace |
| Room Tone Padding | 0.0s / 0.0s | 0.5s Head / 2.0s Tail | Mandatory 0.5s–1.0s / 1.0s–5.0s |
| Format & Bitrate | 24 kHz Mono WAV | 44.1 kHz 320 kbps CBR MP3 | 44.1 kHz ≥ 192 kbps |
BOOKS
The Novella: The Chipped Mug
This audio technology was designed specifically to support an intimate literary story exploring emotional distance, misaligned expectations, and the quiet friction between two people.
2026Julian Brown
The Chipped Mug: Behind the Scenes
The Friction Behind the Fiction
PaperbackKindleAudiobook
Every relationship argument runs on two tracks simultaneously: the words spoken out loud, and the quiet protective calculations running beneath them. Slowing down the tape across all eleven chapters, this companion volume offers an unfiltered, symmetrical autopsy of what was really happening in the room.