IPA Reference
This page summarizes common IPA symbols emitted by hama G2P models and what they represent.
The decoder inventory is sourced from assets/g2p_vocab.json in
/Users/seongmin/hama and /Users/seongmin/hama-training.
How to read this table
- Symbol: IPA token in output text.
- Type: Vowel, consonant, diphthong, or suprasegmental marker.
- Description: Practical articulation label used in linguistic references.
Core vowels
| Symbol | Type | Description |
|---|
i | Vowel | Close front unrounded vowel. |
ɪ | Vowel | Near-close near-front unrounded vowel. |
e | Vowel | Close-mid front unrounded vowel. |
ɛ | Vowel | Open-mid front unrounded vowel. |
a, ɑ | Vowel | Open front/back unrounded vowels. |
ə | Vowel | Schwa (mid central vowel). |
ʌ | Vowel | Open-mid back unrounded vowel. |
ɯ | Vowel | Close back unrounded vowel. |
u | Vowel | Close back rounded vowel. |
ʊ | Vowel | Near-close near-back rounded vowel. |
o, ɔ | Vowel | Close-mid/open-mid back rounded vowels. |
ɝ, ɚ | Vowel | Rhotic vowels (r-colored). |
Frequent consonants
| Symbol | Type | Description |
|---|
p b t d k g | Consonant | Plosives (stop consonants). |
m n ŋ | Consonant | Nasals. |
f v s z ʃ ʒ θ ð h | Consonant | Fricatives. |
t͡ʃ d͡ʒ | Consonant | Affricates (English-style). |
ɹ r ɾ l | Consonant | Liquids/taps/rhotics (language dependent). |
j w | Consonant | Approximants/glides. |
Korean stop and affricate contrasts
| Symbol | Type | Description |
|---|
k, t, p, t͡ɕ | Lenis | Plain/lenis series in Korean. |
kʰ, tʰ, pʰ, t͡ɕʰ | Aspirated | Aspirated series. |
k͈, t͈, p͈, s͈, t͡ɕ͈ | Tense | Tense/fortis series (double articulation tension). |
Common diphthongs
| Symbol | Type | Description |
|---|
aɪ | Diphthong | As in English “price”. |
aʊ | Diphthong | As in English “mouth”. |
ɔɪ | Diphthong | As in English “choice”. |
oʊ | Diphthong | As in English “goat”. |
eɪ | Diphthong | As in English “face”. |
Markers and non-phoneme tokens
| Token | Role | Notes |
|---|
<pad>, <sos>, <eos>, <unk> | Control tokens | Used internally by decoding; not linguistic IPA phones. |
, ,, %, ~ | Formatting markers | May appear in model output depending on text and segmentation. |
¹, ², ³ | Tone/suprasegment | Used for tonal distinctions in relevant languages. |
For the complete decoder set, inspect decoder in
assets/g2p_vocab.json. Treat this page as a practical quick reference for
symbols that appear frequently in downstream applications.
