Pronunciation Coach

Pronunciation Coach uses the concept of pronunciation models to show you how to accurately pronounce any sound or word. Click on the image below to watch a short video on how to create pronunciation models, or try the free web app.

Each pronunciation model contains:

1. Audio, for listening to the pronunciation.
2. A video of the mouth and lips.
3. A video of the speech anatomy.
4. A video showing tongue contact with the roof of the mouth.
5. Phonetic symbols, to identify the individual speech sounds.
6. Waveforms, for viewing timing, pitch and speech intensity (loudness).

Pronunciation dictionary

The Pronunciation Dictionary lets you quickly create pronunciation models from over 21,000 words and sounds. The words and sounds are conveniently arranged in alphabetical, vowel and consonant order, so you can easily locate the words or sounds you want to practise. You can also create custom word lists.

Pronunciation Coach lets you record your speech so you can compare it with the pronunciation model. Click on the image below to watch a short video on how to record and score your pronunciation.

Each recording contains:

1. Audio, for listening to your pronunciation.
2. A video of your mouth and lips (this option requires a webcam).
3. Speech recognition result.
4. Speech intelligibility score.
5. Waveforms, for viewing your timing, pitch and speech intensity (loudness).

Speech recognition

Speech recognition provides feedback on your pronunciation. If your pronunciation is recognized correctly, a green tick and an intelligibility score is displayed. The intelligibility score provides a measure of how well a particular word was pronounced.

If your pronunciation is not recognized correctly, a red cross is displayed.

Once you've learned how to pronounce individual sounds and words, you can use the speech intelligibility scorer to provide feedback on your conversational speech. Click on the image below to watch a short video on how to score your speech intelligibility.

Speech intelligibility refers to the 'understandability' of speech, and the ability to use speech to communicate effectively in everyday situations. It can be affected by many factors, including pronunciation accuracy, audibility and speaking rate.

How it works

The speech intelligibility scorer uses state-of-the-art speech recognition to provide a measure of:

• Speech intelligibility (expressed as a percentage).
• Audibility (loudness).
• Speech rate (number of words spoken per minute).

To score your speech intelligibility, simply enter any text and read it out loud. As you speak, the recognised words are scored from 0 to 100% and highlighted in green. Any unrecognised words are highlighted in red. Once all of the words have been highlighted, the following results are displayed:

Speech intelligibility score

Speech intelligibility is a measure of how easily speech can be understood and is expressed as a percentage. The speech intelligibility scores are categorised as:

Intelligibility description	Intelligibility score
Unintelligible	0%
Poor	10% - 30%
Fair	30% - 50%
Good	50% - 75%
Excellent	75% - 100%

Audibility

In order to understand speech it must be audible. Audibility depends on the loudness of speech and is measured in decibels (dB). The human hearing range goes from 0 dB to 120-140 dB:

Condition	Audibility (dB)
Hearing threshold	0 dB
Whisper	30 dB
Normal conversation	60 dB
Shout	90 dB
Pain threshold	120 -140 dB

Speech rate

Speech rate is the term given to the speed at which we speak and is expressed as the number of words spoken per minute (WPM). Depending on the type of speech, the rate may vary from 100 WPM to over 400 WPM:

Type of speech	Speech rate (WPM)
Slow speech	100 - 110 WPM
Conversational speech	120 - 200 WPM
Audiobooks and podcasts	150 - 160 WPM
Auctioneers and commentators	250 - 400 WPM