A short, first-person audio introduction to my research in philosophy, mathematics, physics, cognitive science, and computational systems, narrated in a synthetic version of my own voice. It is meant as a compact spoken introduction rather than an exhaustive treatment of any one framework. It also documents a practical experiment in producing a recognizable narration from a short voice sample, so that research can be presented in a consistent voice without a complete manual recording.
My research investigates how distinctions arise, persist, transform, and become admissible within systems whose histories cannot simply be reversed. Rather than treating objects, identities, observations, and decisions as independently given, I examine the constraints and processes through which they become distinguishable and operationally meaningful. The recording introduces several interconnected research programmes.
RSVP (Relativistic Scalar-Vector-Entropy Plenum) investigates a constraint-first account of physical reality using scalar capacity, vector flow, and entropy. It explores the emergence of physical structure and temporal ordering through irreversible transformation.
Spherepop Calculus and Spherepop OS develop a computational framework organized around admissible state transitions, provenance, verification, refusal, and irreversible histories. Identity is treated as something established through a history of transformations rather than a static label.
MEM|8 explores reconstructive memory through resonance, interference, phase relationships, and contextual constraints. It investigates retrieval as an active reconstruction process rather than straightforward access to stored representations.
CLIO, HYDRA, and TARTAN address structured reasoning, concurrent scheduling, and locally coordinated computation, respectively.
Admissibility and Irreversibility investigate the distinction between what can be described, constructed, verified, and legitimately committed as an action, particularly under uncertainty and conditions of limited reversibility.
Distinction, Continuity, and Reconstruction examine how identity persists through change, how representations conceal discontinuities, and why successful reproduction does not necessarily establish preservation of the original state or capability.
These frameworks are ongoing research programmes rather than completed or empirically established theories.
The narration was generated in the Hume AI Text-to-Speech Playground from a prepared script, using a custom voice built from a short sample of my speech. The result closely resembles my own voice, although its delivery is slightly more restrained than my ordinary speaking style.
The generated audio was exported as an MP3 and edited by hand. The final word of the narration, "now", was trimmed off the end, and the sound "ile" was cut out of the spoken form of MEM|8 on two occasions using Audacity. The pronunciation of some of the mathematical notation was also adjusted by hand. The final audio is approximately five minutes and forty-one seconds long.
The narration is not a flawless reproduction. Some glitches and false starts from the synthesis remain in the audio, and the voice pronounces "maturity" with a t sound where my own speech uses a ch sound. I have left these as they are. The captions and transcript show what is actually spoken, including the false start "I know this as of how much" just before "CLIO organizes operations" and a repeated "for" in the MEM|8 introduction. The file my-research-frameworks-script.txt records what was originally written.
The voice is represented here by one short audio file, HumeAI_voice-preview_Flyxion.webm, which I downloaded from Hume's interface as a sample. It runs about twenty-five seconds and contains four sentences. It sounds very close to my own speech, including the wider intonation of each sentence, and I cannot say for certain whether it is my original input recording or a preview that Hume generated from it. A Whisper transcription of the file reads as follows.
We just got a new coffee maker and it makes the most delicious coffee ever. I know how hard mornings can be when you do not have a good coffee to start. I researched different models for weeks before deciding which one to buy for our kitchen. Now our morning routine is so much more enjoyable, and we save money too.
Either way, the voice came from a very small amount of material, and that was enough to produce a narration that is recognizably my voice. It was not enough to reproduce the range of my speech. The delivery is more restrained than my own, and the result follows the one manner of speaking present in the sample. I can speak in many different voices and registers, and a faithful model of that range would need far more varied recordings than this one. The sample is published with the project so that the experiment can be judged accurately, including the fact that a short recording already yields a usable likeness.
At the time of this experiment (November 2026), Hume AI's Text-to-Speech Playground offered a free monthly allowance of 10,000 characters, roughly ten minutes of generated speech depending on the length of the text and the speaking rate. The narration of My Research Frameworks used 4,961 characters, about half of the allowance, and the experiment required no paid subscription.
The final audio is my-research-frameworks.mp3, accompanied by the caption file my-research-frameworks.vtt and the plain transcript my-research-frameworks.txt. The original script is my-research-frameworks-script.txt. The voice sample is HumeAI_voice-preview_Flyxion.webm, with its Whisper transcription in the matching .txt and .vtt files.
Flyxion
Independent Researcher