Saystep KoreanA cup of coffee, please.
A cup of coffee, please.
Use listening and imitation to compare ㄱ, ㅋ and ㄲ. Airflow and pitch both matter; English spelling is only a rough aid.
Plain, aspirated and tense › 01 ㄱ
ㄱ
- Listen to the Korean reference once.
- Say the phrase slowly, keeping the meaning in mind.
- Record it again and compare your two recordings.
The tongue, lip and jaw descriptions are practice hints. They do not determine your actual mouth position from the audio alone.
You can practice with synthetic reference audio of the same Korean sentence.
Read aloud with your device's voice
Up to 20 seconds · not too close to the microphone
Microphone recordings are saved on this device. Sign in to get AI feedback.
How the sounds link in this sentence
Practice the places in this sentence where words link together. Listen to the reference first, then say just the one spot you want to work on. What you see here are practice suggestions picked from the Korean sentence.
Linking spots are found from the spelling, on this device. A text match is not a diagnosis of your pronunciation.
These are examples of how words are pronounced in conversation. This is not a diagnosis that the same thing happened in your recording. The synthetic reference voice may pronounce things differently from what is described here.
See the rise and fall of your voice
The free sample is a synthetic voice reading the same Korean sentence. It is not a native speaker in natural conversation.
There is no reference audio, so no comparison curve is shown. Add one with "Overlay the free sample" above.
Every word in the practice sentence (This take)
Tap a word that has a timing to play it. Words that were not detected or have no timing are listed here too. If the chart is wide, swipe sideways to see the rest of the words.
Each voice is shown with its median at 0 semitones. Overall voice height is normalized, and the size of pitch movement is kept. The range shown is ±12 semitones. The line breaks where no pitch could be measured.
Shown in real seconds. Add reference audio or a Before to line them up by the start and end of speech, or by word. We do not score naturalness from curve match or pitch range alone.
Measured values, with each recording's median set to 0. The line breaks where there is no voicing. How closely the curves match is not used as a pronunciation score.
Free: pitch calculation, overlay, manual word sync and playback comparison. Automatic word timings from Azure reuse your existing sound analysis. Use reference audio of the same Korean sentence, said with the same intent.
Sound assessment (for reference)
After you record, you can get a sound assessment.
Pitch and lessons work without a connection. There is no score yet.