g2p-lt-byt5-small — Lithuanian grapheme-to-phoneme (G2P), lite

A byte-level ByT5 model that converts Lithuanian words in standard orthography into their phoneme sequence. This is the lite variant — 300 M params, ~2× faster than the flagship, for edge/embedded use — at a small accuracy cost.

• Input: one word, standard Lithuanian spelling, lowercase — e.g. labas
• Output: space-separated phonemes in the LIEPA-3 phon alphabet — e.g. l a b a s
• Architecture: google/byt5-small (300 M), fully fine-tuned. Byte-level ⇒ no tokenizer issues with Lithuanian letters or phoneme symbols.

For maximum accuracy use the flagship svogunas/g2p-lt-byt5-base.

Results

Held-out test set of 3,000 words unseen in training (true generalisation):

decoding	PER	word accuracy
greedy	1.98 %	88.3 %
beam = 5	1.96 %	88.5 %

The flagship base scores 1.88 % / 89.2 % — a 0.1 % PER edge for ~2× the parameters. This small model was trained for 12 epochs (vs 5 for base): the extra epochs help the small model (2.34 % → 1.98 %) where they did not help base or large.

Usage

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tok = AutoTokenizer.from_pretrained("svogunas/g2p-lt-byt5-small")
model = AutoModelForSeq2SeqLM.from_pretrained("svogunas/g2p-lt-byt5-small")

def g2p(word: str) -> str:
    ids = tok(word.lower(), return_tensors="pt").input_ids
    out = model.generate(ids, max_length=64)          # greedy; add num_beams=5 for a tiny gain
    return tok.decode(out[0], skip_special_tokens=True)

print(g2p("labas"))            # l a b a s
print(g2p("abaravičiene"))     # a b a r a v' i tS' ie n' e

Feed one word at a time (the model is trained on isolated words). For a sentence, split on whitespace, look words up, and handle punctuation/numbers in your own front-end.

Phoneme notation (LIEPA-3 phon alphabet)

Phonemes are space-separated. Conventions:

• Lowercase base symbols for the core sounds.
• ' after a consonant marks palatalization (Lithuanian minkštumas): v', tS', n'.
• Doubled symbols mark length: aa, oo, ii, ee, uu.
• Digraphs for diphthongs/affricates: ie, uo, ea; tS (č), dZ (dž), ts (c), dz.
• Capitalised variants encode stress / length distinctions (e.g. aA, Aa, Oo, Ea).

Full inventory (~90 symbols) observed in the training lexicon:

a i e s s' n' t' t u oo k m r' n r k' l' p j j' m' I v' p' ee d d' l ii S' E v w
g' g uu U aA o S b' tS' b A Aa Ee Uo Z' aa Oo Ii Ea ea ie z J ts' eA z' Z uo O iI
W uU f' dZ' f oO N iE eE N' Uu Ie R' R uO x h L' ts M' x' h' L M tS dZ dz' dz

Training data

• LIEPA-3 word+phoneme TextGrids (phon/) + EMO (Lithuanian emotional speech) phoneme annotations, merged into a 233 k-word pronunciation lexicon (one consistent notation, 99.5 % mean cross-source agreement); 3 k words held out for the test above.
• Excluded by design: medical (different SAMPA notation) and dialectal (phonetic transcripts).

Fine-tuned from google/byt5-small, 12 epochs, lr 5e-4, load_best_model_at_end on eval loss.

Limitations

• Isolated-word model: no sentence-level context, sandhi, or homograph disambiguation.
• Trained on standard Lithuanian; dialectal pronunciation is out of scope.

License & attribution

Released under CC BY 4.0. Training data (LIEPA-3, EMO) is CC BY 4.0 from VDU / CLARIN-LT — please attribute them when you use this model:

Pronunciation lexicon derived from LIEPA-3 and the VDU Lithuanian emotional speech corpus, distributed via CLARIN-LT (CC BY 4.0).

Citation

@misc{g2p-lt-byt5-small,
  title  = {g2p-lt-byt5-small: Lithuanian grapheme-to-phoneme (ByT5, lite)},
  author = {Smaliukas, Arūnas},
  year   = {2026},
  note   = {Fine-tuned from google/byt5-small on a LIEPA-3 + EMO pronunciation lexicon (CC BY 4.0)},
  url    = {https://huggingface.co/svogunas/g2p-lt-byt5-small}
}