Recording analysis & comparison
Examine difficult recordings or compare repeated tests. Scope can include processing outputs, diagnostic plots and an interpretation report that defines its metrics and states what remains unresolved.
PhonoAxis
Biosignal & Acoustic Engineering
Build with sound.
Learn from signals.
Audio and acoustic analysis, reproducible signal-processing workflows and compact prototypes. Cardiovascular signal research runs in parallel.
Engineering services
A useful workflow
around your signal.
Projects begin with a concrete goal: examine a difficult recording, compare acoustic conditions, develop a small model or automate an analysis.
Signal-processing workflows
Turn a repeatable analysis into Python or MATLAB tools: batch processing, feature extraction, event rules, documented outputs and a reproducible handover.
Compact model prototypes
Prepare features, evaluate recorded examples and produce model/export artefacts within an agreed scope. Target-device integration and measurements are separate steps.
Selected founder engineering work
Four difficult recordings examined; a combined report spanning 156 accessory combinations; and an offline audio-to-motion prototype. These are project scopes from my engineering portfolio.
Cardiovascular research direction
Heart sounds carry
timing and structure.
In a patient-grouped evaluation, expert-reviewed cardiac-phase boundaries carried useful information for murmur characterisation. The open problem is obtaining that timing automatically from the heart sound itself. The diagram shows the route tested so far: a research concept, not a diagnostic result.
01 Physiological signal
A beating heart
Heart sounds and murmur patterns within a noisy acoustic signal.
02 Acoustic capture
One sound sensor
A digital heart-sound recording becomes the pipeline’s input.
03 Timing reference
Infer timing: the step under test
The tested route reconstructs an ECG-like timing reference from sound. In October 2026 it did not establish an advantage over equal-time windows.
04 Research output
Murmur-characterisation targets
Presence, intensity grade, timing and shape: the targets studied in this research.
The acoustic signal combines heart sounds, possible murmur patterns and background noise. The timing of heart sounds and murmurs within each beat carries information.
Current finding · October 2026: The tested synthetic-ECG timing route did not establish an advantage over equal-time windows. Whether an automatic timing method can preserve phase information for unseen patients remains open; any clinical use would require separate clinical validation.
The underlying research is my academic work, an MS thesis at NUST and co-authored journal manuscripts, and remains distinct from PhonoAxis ownership and any institutional partnership.
A conceptual view
Sound within
the cardiac cycle.
Explore how a timing reference can divide a heart-sound trace into estimated phases. These teaching traces illustrate the idea; they do not demonstrate model performance.
Heart sound PCGTiming reference Idealised ECG
S1SystoleS2Diastole
An accurate timing reference locates each sound within the cardiac phases. This alignment is idealised; obtaining it automatically is the open research question.
Illustrative traces · not patient dataCurrent stage · October 2026
Work to show.
Questions still to test.
Implemented workflows
Founder experience in audio analysis, recording-investigation reports, analysis scripting and offline prototypes. The portfolio also describes a compact-classifier project workflow. Explore the case studies or inspect the original synthetic example.
Thesis, papers & presentations
An MS thesis, two published first-author conference papers and two ESC Congress presentations. The two first-author journal manuscripts remain under revision (status last checked 9 October 2026). Publication record and status.
Evidence before a product
Test whether automatic cardiac timing can preserve the value of reliable phase information across patients and devices; only then define an evidence-supported workflow with research and clinical collaborators.
Cardiac work is at the research stage. No clinical approval, commercial cardiac deployment or prospective validation is claimed. Venture overview · TXT
Proposed next steps
Define the task.
Test the value.
Engineering delivery and cardiac research have distinct milestones. Each step follows an agreed task or an evidence-based decision.
Engineering delivery
Define the task
Identify the recurring problem and agree the intended user, inputs and output.
Agree how success is assessed
Set acceptance criteria, recording assumptions, platform constraints and sharing boundaries.
Deliver an inspectable workflow
Provide the agreed outputs, reproducible methods and a concise account of limitations.
Identify repeatable work
Assess which demonstrated workflow could become a maintained tool.
Cardiac research gates
Find a reliable automatic timing route
The first tested route did not establish an advantage over equal-time windows. Test whether an automatic timing method can preserve the value of reliable phase boundaries for unseen patients, under the same patient-grouped comparison.
Evaluate transfer and uncertainty
Test the complete workflow across recording systems and cohorts, including failure cases.
Define a supported application
Identify a useful research workflow with domain collaborators and establish the additional validation it needs.
Founder
Zafar Ahmad.
I am an electrical engineer and cardiovascular signal-processing researcher with an MS from the National University of Sciences and Technology (NUST). My work spans clinical acoustic acquisition, deep learning, wearable prototypes and audio engineering.
My approach starts with the recording and a reproducible baseline, then adds models where the task benefits from them. Each delivery includes the outputs, supporting analysis and the limits of the result.
Bring a question.
Let’s start with the signal.
For an engineering enquiry, describe the goal, input format and approximate duration or volume, acquisition conditions, required outputs and any target platform or deadline. Please describe the task before sending any recordings.
For a research collaboration, introduce the cardiac measurement, timing-validation or transfer question you want to investigate.
ahmadzafar577@gmail.com