More
Most numbers here are worked out on your phone with simple measurements of loudness, pitch and hiss. They are useful for spotting habits and tracking change, not a clinical assessment. Here is what each one does, and what it can't do.
Pace
Words in the text divided by the time from your first sound to your last, so pauses count. 140 to 160 words a minute is the default target for a fast talker slowing down; 120 to 150 is typical conversation. Background noise at the start or end of a take can nudge it.
Trailing off
For each phrase (the stretch between two pauses of 0.3 s or more), the loudness of the last quarter is compared with the first three quarters. A drop of 6 dB or more counts as trailing off. There is no published clinical threshold for this, so treat it as a rule of thumb. It only works because the app turns off your phone's automatic volume boost.
Expressiveness
The spread of your pitch across a take, in semitones. Under 3 tends to sound flat. It is a side note, not one of the headline numbers.
Word stress
Compares the loudness, pitch and length of the two syllables. When the difference is too small to be sure, it says it couldn't tell instead of guessing.
Sound scores (with Azure)
Microsoft's pronunciation assessment scores each word and sound against its own model of the accent. It is a model, not a listener: a low score means it was less sure it heard the target, not that a person would misunderstand you. With British scoring (en-GB) you are judged against a British model, but you get less detail: no prosody score, and sound names can be missing. American scoring (en-US) gives more detail but can mark a correct British vowel down.
Without an Azure key
You listen to your take next to the model and rate it yourself. Those ratings decide when an item comes back for review, but they never count towards your sound scores, so the numbers stay honest.
The "s" in S-Lab
The dial shows the centre of your hiss. A clear "s" sits high, a slushy one drifts down towards "sh", and calibrating fits the target to your mouth and microphone. It hears the colour of the sound, not your tongue, so it can't detect a dentalised or interdental lisp (an "s" that sounds like "th"). Always listen back as well.
When to see a speech therapist
Drilling helps once you know the target. If you don't, a trained eye usually gets you there faster. Book a speech-language therapist if:
- your "s" hasn't moved after two weeks of regular S-Lab practice (Today will tell you when this happens),
- a sound feels physically impossible however you try it,
- you stammer, or speaking feels like a struggle,
- people keep asking you to repeat yourself even after weeks of practice.
One or two sessions are often enough to learn the tongue position. Then keep using Enunciate for the daily practice. If your voice is hoarse or sore for more than three weeks, see a doctor first.