A Lingala automatic speech recognition (ASR) model, further fine-tuned from keystats/w2v-bert-2.0-lingala-main-best (itself fine-tuned from facebook/w2v-bert-2.0).

This is a continued fine-tuning run: starting from the main-best checkpoint's weights, training continued for 3 more epochs at a lower learning rate (1e-5 vs. main-best's 3e-5), everything else — data, splits, casing, filtering — kept the same.

Model description

facebook/w2v-bert-2.0 — a large-scale, multilingual self-supervised speech encoder pretrained with a BERT-style masked prediction objective — is used as the backbone, with a character-level CTC (Connectionist Temporal Classification) head. This checkpoint continues fine-tuning from main-best rather than starting from the base model.

Text casing note: training targets were kept in their raw, cased form (same tokenizer/vocab as main-best).

Training data

Same pool and split usage as main-best — WAXAL validation held out, test untouched:

Source	Role
google/WaxalNLP (`lin_asr` config)	`train` split pooled into training, `validation` split held out untouched as the fixed evaluation benchmark
KasuleTrevor/Lingala_100hrs	All splits pooled into training

WAXAL's validation split is the only data used for evaluation, and it was never included in training.

Training procedure

Base checkpoint: keystats/w2v-bert-2.0-lingala-main-best (continued fine-tuning, not from facebook/w2v-bert-2.0 directly)
Architecture: Wav2Vec2BertForCTC, add_adapter=True
Processor: same Wav2Vec2BertProcessor / tokenizer as main-best (character-level, raw/cased text, | word delimiter, [PAD] as CTC blank)
Sample rate: 16 kHz mono
Epochs: 3 additional epochs on top of main-best
Learning rate: 1e-5 (lower than main-best's 3e-5, cosine schedule, 10% warmup)
Effective batch size: 32 (per-device batch size 4 × gradient accumulation 8)
Precision: fp16, gradient checkpointing enabled
Regularization: attention/hidden/feature-projection dropout 0.05
Data filtering: same CTC-feasibility filtering as main-best
Seed: 42 (deterministic — same seed for Python/NumPy/PyTorch/CUDA)

Evaluation results

Evaluated on the WAXAL Lingala validation split, greedy decoding vs. greedy + KenLM (keystats/waxal-kenlm-models-best). Adding the KLM gives a consistent, meaningful WER/CER improvement over greedy decoding alone — pair the two for the best results.

How to use

Option 1 — model alone (greedy decoding)

python

import torch
import librosa
from transformers import Wav2Vec2BertForCTC, Wav2Vec2BertProcessor

MODEL_ID = "keystats/w2v-bert-2.0-lingala-main-best-3"
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"

processor = Wav2Vec2BertProcessor.from_pretrained(MODEL_ID)
model = Wav2Vec2BertForCTC.from_pretrained(MODEL_ID).to(DEVICE).eval()

audio_array, sr = librosa.load("path/to/audio.wav", sr=16000, mono=True)
inputs = processor(audio_array, sampling_rate=16000, return_tensors="pt")
with torch.no_grad():
    logits = model(input_features=inputs.input_features.to(DEVICE)).logits

predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)[0]

print(transcription)  # cased, punctuated Lingala text

Option 2 — model + KLM (recommended, higher accuracy)

python

# pip install pyctcdecode
# pip install https://github.com/kpu/kenlm/archive/master.zip

import torch
import librosa
from huggingface_hub import hf_hub_download
from transformers import Wav2Vec2BertForCTC, Wav2Vec2BertProcessor
from pyctcdecode import build_ctcdecoder

MODEL_ID = "keystats/w2v-bert-2.0-lingala-main-best-3"
KLM_REPO_ID = "keystats/waxal-kenlm-models-best"
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"

processor = Wav2Vec2BertProcessor.from_pretrained(MODEL_ID)
model = Wav2Vec2BertForCTC.from_pretrained(MODEL_ID).to(DEVICE).eval()

klm_path = hf_hub_download(repo_id=KLM_REPO_ID, repo_type="dataset",
                            filename="lingala/lingala_5gram_correct-best.arpa")

def build_vocab_list(tokenizer, vocab_size):
    vocab_dict = tokenizer.get_vocab()
    vocab_list = [None] * vocab_size
    for tok, idx in sorted(vocab_dict.items(), key=lambda kv: kv[1]):
        if idx < vocab_size:
            vocab_list[idx] = tok
    pad_id = tokenizer.pad_token_id
    if pad_id is not None and pad_id < len(vocab_list):
        vocab_list[pad_id] = ""
    word_delim = getattr(tokenizer, "word_delimiter_token", None)
    if word_delim:
        delim_id = vocab_dict.get(word_delim)
        if delim_id is not None:
            vocab_list[delim_id] = " "
    return vocab_list

vocab_list = build_vocab_list(processor.tokenizer, model.config.vocab_size)
decoder = build_ctcdecoder(
    vocab_list,
    kenlm_model_path=klm_path,
    alpha=0.5,
    beta=0.7,
)

audio_array, sr = librosa.load("path/to/audio.wav", sr=16000, mono=True)
inputs = processor(audio_array, sampling_rate=16000, return_tensors="pt")
with torch.no_grad():
    logits = model(input_features=inputs.input_features.to(DEVICE)).logits

transcription = decoder.decode(logits.cpu().numpy()[0], beam_width=100)
print(transcription)

Intended uses & limitations

Intended for transcribing spoken Lingala audio into cased, punctuated text.
As a CTC-based model, it assumes single-speaker, forward-only audio and has no mechanism for overlapping speech from multiple speakers.
This is a continued fine-tune of main-best; if main-best had already converged, additional low-LR epochs may yield only marginal gains (or overfitting) — compare validation metrics against main-best before choosing between them.

Citation

bibtex

@misc{keystats_wav2vec2bert_lingala_best3,
  title={w2v-bert-2.0-lingala-main-best-3: A Lingala ASR model, continued fine-tune of w2v-bert-2.0-lingala-main-best},
  author={keystats},
  year={2026},
  howpublished={\url{https://huggingface.co/keystats/w2v-bert-2.0-lingala-main-best-3}}
}

@misc{waxal,
  title={WAXAL: A Multilingual African Speech Dataset},
  author={Google},
  howpublished={\url{https://huggingface.co/datasets/google/WaxalNLP}}
}

@misc{kasule_lingala_100hrs,
  title={Lingala\_100hrs},
  author={KasuleTrevor},
  howpublished={\url{https://huggingface.co/datasets/KasuleTrevor/Lingala_100hrs}}
}

@inproceedings{w2vbert2,
  title={Seamless: Multilingual Expressive and Streaming Speech Translation},
  author={Seamless Communication and others},
  year={2023},
  howpublished={\url{https://huggingface.co/facebook/w2v-bert-2.0}}
}