When we speak, we subconsciously listen to vowel and consonant sounds to ensure we're producing them correctly. If our speech is slightly different from what we intended, we adjust the way we speak to correct for these slight errors.
For some of us, detecting these errors or making the necessary adjustments can be a challenge. That's why we developed the icSpeech Analyzer.
How does the speech analyzer work?
icSpeech Analyzer uses the clinically proven technique of biofeedback to help improve your speech. Biofeedback lets you see your speech on a computer screen in real-time. At first, you learn to control your speech while looking at the display and comparing your speech to an example. Then, with practise, you learn to control your speech without looking at the display.
icSpeech Analyzer supports the following biofeedback displays:
You can also record your speech, create target displays, and perform measurements on a range of speech parameters.
Waveform displays are useful for controlling speech rate, loudness and pitch. They also show the difference between voiced consonants, voiceless consonants and vowel sounds.
Waveforms scan across the display from right to left, with intensity or frequency plotted vertically and time horizontally. The time taken to scan the entire display can be adjusted to suit your needs.
The above example shows a speech waveform (top) and a voice intensity waveform (bottom). This example clearly illustrates the difference between the voiced sound /b/ and the unvoiced sound /p/.
You can also save the waveform display and perform precise measurements on a range of speech parameters, including:
- Speech rate (independent of pauses or dysfluencies).
- Speech intensity (loudness).
- Voice pitch.
- Voice intensity.
- Voice onset time.
- Sentence, word, syllable and phoneme duration.
The real-time bar is a useful biofeedback tool that shows a simple bar, which moves upwards with increased pitch, loudness or voice intensity. This allows you to monitor your speech, make corrections, and instantly see the result. Bar charts are particularly useful when working with sustained sounds.
The above example shows pitch (yellow bar) and a target (blue bar). Targets are adjustable bars that appear next to the active bar. Depending on the task, you can try to keep below, reach or exceed the target.
Mirrors are often used in speech therapy as they allow you to view lips, teeth, jaw, tongue and facial muscles during speech.
Visual feedback, using the built-in webcam feature, takes this a step further. With a webcam you can:
- Pause your image for closer inspection.
- Record your video and audio.
- Practise your speech whilst watching example recordings.
A spectrogram is a visual representation of sound. It scans across the display from right to left, with frequency plotted vertically and time horizontally. The intensity of sound, or loudness, is indicated by grey scale or color intensity. The spectrogram below clearly shows the difference between the /r/ sound and the /w/ sound.
Studies have shown that some people, especially children, have difficulty detecting when they misproduce certain sounds. Spectrograms provide additional feedback, helping you to match your speech to a visual target. At first, you rely on visual feedback and learn to ignore your auditory feedback (people often say their correct speech production sounds wrong). Once you've mastered the correct speech production, you can start to rely more on your auditory feedback and less on the visual feedback.
Spectrograms can be particularly useful for people that substitute the following sounds:
- /w/ for /r/
- /w/ for /l/
- /m/ for /b/
- /p/ for /b/
- /s/ for “sh” or vice versa
The spectrum display provides an alternative method of visually representing the acoustic signal. The display is ideal for viewing fundamental frequency (pitch), harmonic frequencies, and formant frequencies (vocal tract resonances). Speech intensity, or loudness, is plotted vertically and frequency horizontally.
The above displays illustrate the frequency spectrum for the vowel sound /i/. The top display (FFT) shows the fine spectral detail associated with the fundamental frequency and its harmonics. The bottom display (LPC), highlights the formant frequencies by removing the fundamental and harmonic frequencies.
|Operating Systems||Windows 10, 8, 7, Vista and XP|
|CPU||Minimum 1 GHz Intel® Pentium® or equivalent|
|Memory||Minimum 1 GB of RAM|
|Storage||100 MB of free hard disk space
|Microphone||Any PC compatible microphone|
|Webcam (optional)||Any USB Windows compatible webcam|