Faster TOEFL Speaking score estimates. Clearer practice decisions. Smarter preparation.

Today, My Speaking Score is introducing the Vox AI Scoring Engine, our proprietary scoring system for TOEFL Speaking practice.

Vox was built for one purpose: to help TOEFL Speaking students understand their performance faster, more clearly, and more affordably.

For years, My Speaking Score has helped test takers use data-powered feedback to prepare for TOEFL Speaking. Vox takes that mission forward. It gives learners a fast, consistent score estimate so they can stop guessing and start making better practice decisions.

When students practice speaking, they need answers to simple questions:

Was this response strong enough?
Did I improve?
Am I getting closer to my target score?
Which responses are stronger or weaker?

Vox helps answer those questions.

What is Vox?

Vox is My Speaking Score’s proprietary AI scoring engine for spoken English responses.

It analyzes a student’s spoken answer and produces a score estimate that reflects overall TOEFL Speaking performance. Vox combines speech recognition, audio analysis, and machine learning models to evaluate spoken responses at scale.

Vox is not just speech-to-text. Speech-to-text converts audio into words. Vox evaluates speaking performance.

Component	What it does
Speech-to-text	Converts spoken audio into written text.
Audio analysis	Evaluates speech patterns such as pacing, delivery, and fluency signals.
Machine learning scoring	Uses performance data to estimate speaking proficiency.
Vox AI Scoring Engine	Combines these signals to estimate TOEFL Speaking performance.

The result is a faster feedback loop for TOEFL Speaking practice.

Why Vox matters now

Most TOEFL Speaking students practice without a clear measurement system.

They record responses. They listen back. They may ask a teacher, a tutor, or a friend for feedback. But they often still do not know whether their score is improving.

That creates uncertainty. And uncertainty slows progress.

Vox gives students a clearer signal. It helps them measure practice attempts, compare responses, and track progress over time.

Student problem	How Vox helps
“I do not know if I am improving.”	Vox gives a score estimate after practice.
“I cannot tell which response was better.”	Vox helps compare stronger and weaker responses.
“I do not know if I am close to my goal.”	Vox helps learners track movement toward a target score.
“I need feedback faster.”	Vox supports immediate, scalable score estimation.

This is the core of data-powered TOEFL Speaking prep: practice, measure, compare, adjust.

How Vox performed in an initial validation sample

To evaluate Vox, we compared Vox score estimates with scores from a licensed ETS scoring engine across an initial validation sample of 205 spoken responses.

The results showed strong alignment.

Metric	Result	What it means
Sample size	205 spoken responses	A meaningful initial sample for evaluating whether the two scoring systems move together.
Mean signed score difference	0.04 / 5	Vox showed minimal average scoring bias relative to the licensed ETS scoring engine.
Pearson correlation	0.784	Vox scores showed strong overall alignment with the licensed ETS scoring engine.
Spearman rank correlation	0.841	Vox was especially strong at ranking stronger and weaker performances in a similar order.

In plain English: Vox and the licensed ETS scoring engine usually move in the same direction.

When the licensed ETS scoring engine gives a higher score, Vox usually gives a higher score too. When the licensed ETS scoring engine gives a lower score, Vox usually does the same.

The two systems are not identical. Some disagreement on individual responses is expected. But across the sample, Vox tracked speaking performance in a strongly aligned way.

What does a 0.784 correlation mean?

Correlation measures how closely two sets of scores move together.

A Pearson correlation of 0.784 means Vox and the licensed ETS scoring engine showed a strong positive relationship. In practical terms, the two systems tended to recognize stronger and weaker TOEFL Speaking performances in similar ways.

Speaking assessment is complex. A spoken response includes fluency, delivery, pacing, pronunciation, vocabulary, grammar, organization, and idea development. Because human speech is variable, perfect agreement between scoring systems is not expected.

The important question is whether two scoring systems show stable agreement across many responses.

In this initial validation sample, Vox did.

Correlation range	Plain-English meaning	How to understand Vox
0.00 to 0.30	Weak relationship	The systems would not be closely aligned.
0.30 to 0.60	Moderate relationship	The systems would show some agreement, but with more variation.
0.60 to 0.80	Strong relationship	Vox and the licensed ETS scoring engine generally moved together.
0.80+	Very strong relationship	The systems show very strong agreement in overall score movement or ranking.

The key takeaway:

Vox showed strong alignment with a licensed ETS scoring engine in an initial validation sample of 205 spoken responses.

Why ranking agreement matters

For TOEFL Speaking students, progress is often about comparison.

Was today’s response better than yesterday’s? Was this answer stronger than the last one? Is the student moving closer to the target score?

That is why Spearman rank correlation matters.

Spearman correlation measures whether two scoring systems rank performances in a similar order. Vox showed a Spearman rank correlation of 0.841, which means it was especially strong at identifying stronger and weaker responses in a similar way.

Student question	Why ranking agreement matters
“Which response was better?”	Vox helps compare performance across practice attempts.
“Am I improving?”	Vox helps students track movement over time.
“Which responses are closer to my target?”	Vox helps identify stronger performances more consistently.
“Should I change my practice strategy?”	Vox gives learners a data point they can use to make better decisions.

This matters because better practice starts with better measurement.

What Vox means for TOEFL Speaking students

Vox gives students a faster way to understand their practice.

Instead of recording responses and wondering whether they are improving, learners can score, compare, and adjust. That creates a stronger feedback loop.

With Vox, students can:

estimate their TOEFL Speaking performance,
compare practice responses,
identify stronger and weaker attempts,
track progress over time,
and make more confident practice decisions.

This is what My Speaking Score was built to do: help TOEFL Speaking students use data to improve.

What Vox does not mean

Vox is a score estimation engine for practice.

It does not replace the TOEFL test, and it does not guarantee a future TOEFL result. Speaking scores can vary depending on the task, topic, response quality, delivery, language use, and test-day performance.

Vox gives learners a useful practice signal. It helps them understand where they are, how responses compare, and whether their performance is moving in the right direction.

That signal can make preparation clearer, faster, and more focused.

The future of data-powered TOEFL Speaking preparation

My Speaking Score was built around a simple belief: speaking improvement becomes easier when learners can see their performance clearly.

Vox moves that belief forward.

It gives My Speaking Score a proprietary scoring engine that can support faster practice, better feedback, and more scalable learning experiences for TOEFL Speaking students around the world.

The future of speaking preparation will be more immediate, more personalized, and more data-informed.

Vox is part of that future.

‍

FAQ

What is the Vox AI Scoring Engine?

Vox is My Speaking Score’s proprietary AI scoring engine for spoken English responses. It estimates TOEFL Speaking performance so learners can better understand their practice results.

Is Vox just speech-to-text?

No. Speech-to-text converts spoken audio into written text. Vox uses speech recognition, audio analysis, and machine learning models to estimate speaking performance.

Does Vox replace the TOEFL test?

No. Vox is designed for TOEFL Speaking practice. It provides a score estimate that helps learners prepare more effectively.

How accurate is Vox?

In an initial validation sample of 205 spoken responses, Vox showed strong alignment with a licensed ETS scoring engine. Vox had a Pearson correlation of 0.784 and a Spearman rank correlation of 0.841.

What does the 0.04 score difference mean?

The mean signed score difference between Vox and the licensed ETS scoring engine was 0.04 points on a 5-point scale. This means Vox showed minimal average scoring bias across the validation sample.

What does correlation mean in simple terms?

Correlation shows how closely two sets of scores move together. A strong positive correlation means that when one scoring system gives a higher score, the other system usually gives a higher score too.

Why does Spearman correlation matter?

Spearman correlation measures ranking agreement. In this comparison, the Spearman correlation was 0.841, which means Vox was especially strong at identifying which responses were stronger or weaker overall.

Does Vox give the exact same score as the licensed ETS scoring engine every time?

No. The systems are strongly aligned overall, but they are not identical. Some disagreement on individual responses is expected in speaking assessment.

Is 205 responses enough for validation?

A 205-response sample is enough for an initial validation analysis showing whether two scoring systems move together in a statistically meaningful way. Future testing across larger and more diverse samples will help further validate Vox across more speakers, tasks, accents, and proficiency levels.

How should TOEFL students use Vox?

Students should use Vox to estimate performance, compare responses, track progress, and make better practice decisions.

Can Vox predict my exact TOEFL Speaking score?

Vox provides an estimate, not a guarantee. It can help learners understand their likely performance level during practice, but test-day results can vary.

Why did My Speaking Score build Vox?

My Speaking Score built Vox to give TOEFL Speaking learners faster, more scalable, data-powered feedback. The goal is to help students practice with clearer information and make better decisions as they prepare.

What is the main takeaway?

Vox showed strong alignment with a licensed ETS scoring engine in an initial validation sample. It gives TOEFL Speaking learners a practical way to estimate performance, compare responses, and track progress during practice.

‍