English syllables are identified with their IPA.
The content of word_to_ipa_with_syllables.json
{'a': [['a', 'ei']],
'aback': [['a', 'ə'], ['back', 'bæk']],
'abandon': [['a', 'ə'], ['ban', 'bæn'], ['don', 'dn̩']],
'abandoned': [['a', 'ə'], ['ban', 'bæn'], ['doned', 'dn̩d']],
'abandonment': [['a', 'ə'], ['ban', 'bæn'], ['don', 'dn̩'], ['ment', 'mn̩t']],
'abated': [['a', 'ə'], ['ba', 'bei'], ['ted', 'tɪd']],
'abatement': [['a', 'ə'], ['bate', 'beit'], ['ment', 'mn̩t']],
'abet': [['a', 'ə'], ['bet', 'bɛt']],
'abide': [['a', 'ə'], ['bide', 'bɑɪd']],
'abiding': [['a', 'ə'], ['bi', 'bɑɪ'], ['ding', 'dɪŋ']],
'ability': [['a', 'ə'], ['bi', 'bɪ'], ['li', 'lə'], ['ty', 'ti']],
'abiotic': [['a', 'ei'], ['bi', 'bɑɪ'], ['o', 'ɑ'], ['tic', 'ɾɪk']],
'ablaze': [['ab', 'əb'], ['laze', 'leiz']],
'able': [['a', 'ei'], ['ble', 'bl̩']],
'ably': [['ab', 'eib'], ['ly', 'li']],
'aboard': [['a', 'ə'], ['board', 'bɔɹd']],
'abode': [['a', 'ə'], ['bode', 'boʊd']],
'abolish': [['a', 'ə'], ['bo', 'bɑ'], ['lish', 'lɪʃ']],
'abominable': [['a', 'ə'],
['bo', 'bɑ'],
['mi', 'mə'],
['na', 'nə'],
['ble', 'bl̩']],
'abomination': [['a', 'ə'],
['bo', 'bɑ'],
['mi', 'mə'],
['na', 'nei'],
['tion', 'ʃn̩']]
...
}
The results are good enough, but certainly contain errors and ellipsis.
Use machine learning to generate IPA for off-list words