Hvordan HolosAccent gør din stemme til en udtale‑score

Preparing your recording

Først dekodes lyden til et standardformat: mono lyd samplet ved 16 kHz.

Systemet estimerer optagelsens signal‑til‑støj‑forhold. Hvis den estimering falder under 25 dB og støjfiltrering er tilgængelig, anvender det et støjreducerende filter før analysen. Hvis filtreringen fejler, fortsætter det med den originale optagelse.

Waveformen normaliseres derefter for at matche taleanalysemodellens inputkrav. Stilhed og optagelser kortere end et halvt sekund får en nulscore.

Allowing different pronunciations

Derefter konverteres reference‑teksten til fonetiske udtaler ved hjælp af en udtale‑engine. Et ord kan have flere accepterede former.

Udtalesættet inkluderer:

• British and American variants.
• Common reduced forms, such as the unstressed vowel in to.
• Selected contextual alternatives.

For eksempel adskiller den afsluttende konsonant i close the door sig fra den i a close friend. Systemet kan vælge mellem de understøttede alternativer, mens det behandler sætningen.

Listening to sounds, not reconstructing words

Optagelsen går gennem en akustisk model. Den producerer beviser for forskellige talelyde over tid, med ca. 20‑millisekunders intervaller.

Dette er en fonetisk repræsentation af optagelsen. Scoreren bruger ikke en ord‑niveau sprogmodel til at rekonstruere sætningen.

Systemet undersøger derefter de akustiske output på to måder:

Metode	Hvad den gør
Expected-sound alignment	Finder den bedste timing for de forventede lyde og vælger en komplet udtalevariant for hvert ord.
Independent sound recognition	Følger de lyde, som modellen foretrækker, uden at kræve reference‑sætningen.

Begge perspektiver er nødvendige, fordi justering kan placere en forventet lyd et sted i optagelsen, selvom taleren udelod den. Uafhængig lydgenkendelse giver ekstra bevis om, hvad der faktisk blev sagt.

Matchning tager også timing i betragtning, så det er sværere for en lyd i ét ord at tælle som bevis for et fjernt ord.

Scoring each sound

For hver justeret lyd beregner scoreren et pronunciation accuracy measure: hvordan stærkt modellen støtter den forventede lyd i forhold til konkurrerende lyde i det interval.

Den overvejer også:

• Den forventede lyds peak probability.
• Om independent sound recognition fandt en matching sound nearby.

En fitted mathematical mapping omdanner dette bevis til en sound accuracy value between 0 and 100.

En lyd, der mangler i det independent recognition result does not automatically receive zero. The acoustic evidence still matters.

Deciding which words need attention

Word feedback combines these sound values with explicit error checks. A word must:

• Have enough evidence that it was spoken.
• Meet the configured average accuracy threshold.
• Avoid detected substitutions, insertions, or severe missing-sound errors.

Clear substitutions require weak support for the expected sound and strong support for a competing sound.

Evidence	Current threshold
Support for the expected sound	Below 0.05
Support for its replacement	At least 0.5

These are model evidence thresholds, not probabilities that a human listener would judge the word incorrect.

For flagged words, the system returns the expected pronunciation and its estimate of the sounds heard, written in the International Phonetic Alphabet (IPA). That estimate can be imperfect, especially around word boundaries.

Building the sentence score

The sentence score starts with a fitted combination of:

• Average sound accuracy.
• The share of expected sounds matched by independent sound recognition.
• The number of speech sounds pronounced per second.
• Long pauses between words.

The timing calculation excludes silence before the first aligned sound and after the last.

These features contribute to one overall score. The current implementation does not provide separate user-facing fluency, completeness, or intonation scores.

How word credit affects the result

The scorer then applies an explicit word-credit multiplier. Each word is weighted by the number of sounds in its selected pronunciation.

Word result	Credit earned
Correct	Full weight
Spoken but flagged	One quarter of its weight
Omitted	No credit

The calculation is:

word credit =
    (correct sound weight + 0.25 × flagged spoken sound weight)
    / total expected sound weight

raw score =
    sentence estimate, clipped to 0–100, × word credit

Consequently, a long incorrect word usually costs more than a short one. Reading only a small part of the sentence also limits the score, even if that fragment sounds clear.

Comparing your score with the reference

Finally, the system can normalize the raw score using a reference recording of the same phrase.

It scores the synthesized reference with the same scorer version and, when a usable matching baseline is already available, applies:

displayed score = min(100, raw score / reference score × 100)

For example, a raw score of 80 against a reference score of 88 becomes approximately 90.9.

This adjustment accounts for the fact that the model may give even the reference recording less than 100.

Baselines are tied to:

• The reference voice and speech-generation model.
• The scoring model.
• The vocabulary.
• The calibration and rule version.

If a matching baseline is unavailable or unusable, the result keeps its raw score. A first attempt can therefore remain unadjusted while a baseline is prepared for later attempts.

What your score means

The resulting number is a practice score on HolosAccent’s scale.

A score of 90 does not mean that 90 percent of your words were correct, and it does not establish a CEFR level or native-speaker equivalence.

The mapping has been fitted using expert-rated recordings, but the new scoring system remains experimental and its full acceptance criteria have not yet been verified. It can miss errors or flag acceptable speech.

Individual word feedback and patterns across repeated attempts provide context for interpreting the number.