Pitch & Fundamental Frequency
Typical fundamental frequency, usable pitch range, pitch movement and stability may be compared across suitable voiced passages.
Voice & Speaker Comparison
Aesonlabs compares questioned speech with known voice material using controlled listening, acoustic measurements and speech-pattern analysis. The objective is to determine whether the available recordings support, conflict with or are insufficient for a proposed speaker relationship.
Do the Recordings Contain Speech From the Same Speaker?
Speaker comparison is not based on one pitch value, one visible formant or a subjective impression that two voices sound alike. Multiple characteristics are examined across comparable speech, and the strength of any conclusion depends on the quality, duration and suitability of the submitted material.
Have questioned and known voice recordings? Submit Recordings
COMPARISON CHARACTERISTICS
A Voice Is Evaluated Through Several Independent Features
No single characteristic is treated as a voiceprint. Measurements are interpreted together and within the limitations of each recording.
Formants & Vocal Resonance
Formant behaviour and resonance patterns can provide information about vocal-tract configuration when comparable vowels and adequate spectral detail are available.
Intonation & Prosody
Sentence melody, emphasis, rhythm, pauses and recurring timing habits are examined across natural speech rather than isolated words alone.
Pronunciation & Articulation
Consonant production, vowel realization, reductions, word endings and other recurring articulation patterns may support or conflict with a proposed comparison.
Accent, Dialect & Language Use
Accent and dialect features may be relevant, but they are behavioural characteristics that can vary with context, language, audience and deliberate disguise.
Voice Quality & Speaking Style
Breathiness, creak, nasality, vocal effort, speech rate and other qualities are considered while accounting for health, emotion and recording conditions.
ACOUSTIC MEASUREMENT
Formants Contribute Evidence, but They Do Not Stand Alone
Formants are areas of acoustic energy associated with resonance in the vocal tract. Their positions and movement may be examined when the recordings contain comparable vowels and sufficient frequency information.
The same speaker can produce different measurements because of surrounding sounds, emotional state, vocal effort, illness, microphone position, telephone transmission or compression. Different speakers can also share broadly similar measurements.
For that reason, formant observations are correlated with fundamental frequency, pronunciation, timing, intonation, voice quality and the known recording conditions. The purpose is to evaluate the combined pattern of agreement and disagreement—not to assign identity from one visual feature.
APPROPRIATE TEST MATERIAL
The Reliability of the Comparison Begins With the Recordings
Strong source material cannot be replaced by software. Whenever possible, provide the original files and enough natural speech to evaluate recurring characteristics.
Preferred Comparison Material
- Original or least-processed audio files
- Several passages of clear, natural speech
- Similar words, vowels or phonetic content
- Comparable conversational speaking styles
- Known recordings from a reasonably similar period
- Information about each device and recording method
Conditions That May Limit Reliability
- Very short clips or isolated sounds
- Whispered, shouted or deliberately disguised speech
- Heavy music, noise or overlapping speakers
- Severe telephone or messaging-app compression
- Speech recorded at very different distances
- Material that has been repeatedly converted or edited
CONTROLLED COMPARISON
From Submitted Samples to a Documented Finding
Preserve and Inventory
The questioned and known recordings are retained unchanged. File identity, format, duration and available source information are documented.
Assess Sample Suitability
Speech quantity, clarity, channel quality, compression, background interference and comparability are evaluated before detailed measurements are interpreted.
Select Comparable Passages
Relevant speech segments are identified, with preference given to similar phonetic content and comparable speaking conditions.
Compare Acoustic and Speech Features
Controlled listening and appropriate measurements are used to examine agreements, differences and alternative explanations.
Correlate, Document and Report
Observations are considered together and reported using language proportionate to the quality and strength of the available evidence.
RESULT LANGUAGE
The Result May Support, Conflict With or Be Unable to Resolve the Proposition
Conclusions are not reduced to a percentage generated by one piece of software.
A
Features support the proposed speaker relationship. The available agreements may provide support when meaningful unexplained differences are not present.
B
Features conflict with the proposed relationship. Repeated, technically meaningful differences may support exclusion or another qualified adverse finding.
C
The material is insufficient or inconclusive. Limited speech, poor comparability or signal degradation may prevent a reliable determination in either direction.
Controlled Work and Technical Documentation
Original submissions are not overwritten. Working copies, selected comparison passages, relevant measurements, screenshots, observations and limitations are maintained according to the agreed scope. When required, a report can identify the material examined, methodology, findings and the basis for the resulting opinion.
Court attendance, testimony, conferences with counsel and preparation for examination or cross-examination are quoted separately. The legal admissibility and ultimate weight of an opinion remain matters for the court or tribunal.
Voice & Speaker Comparison FAQ
Can you identify a person from a voice recording?
The service compares questioned speech with supplied known material. The strength of any opinion depends on sample quality, quantity and comparability. A recording does not provide an infallible biometric identification by itself.
How much speech is required?
There is no universal minimum. Several clear passages containing varied natural speech are preferable. A short clip may permit only a limited comparison or no reliable conclusion.
Can telephone or voicemail recordings be compared?
Often they can be examined, but telephone bandwidth, compression, transmission errors and device differences can remove or distort useful characteristics. Similar recording conditions are preferred.
Can accent alone establish that two speakers are the same?
No. Accent can support a broader comparison, but it may be shared by many speakers and can change with context, language, location and deliberate imitation or disguise.
Can whispered or disguised speech be compared?
Possibly, but reliability may be substantially reduced because whispering and disguise alter pitch, resonance, articulation and speaking behaviour. Suitable like-for-like known samples are especially important.
Can the comparison be documented for legal use?
Yes. Preservation, examination scope and reporting requirements should be established before work begins. Admissibility remains a legal determination.
Submit the Questioned and Known Recordings for Evaluation
Include the source of each recording, the identity represented by each sample, approximate relevant timestamps and whether formal reporting may be required.
Original or least-processed files are preferred. Open an Audio Case