---
license: agpl-3.0
library_name: pytorch
language:
- hi
- or
- ta
datasets:
- google/fleurs
tags:
- forced-alignment
- phoneme-recognition
- phoneme-segmentation
- speech
- multilingual
---

# mbfa-brahmic (p4mbfa · brahmic)

A context-free CNN phoneme encoder and forced aligner (p4mbfa, the successor of CUPE / p3cupe),
for the **brahmic** language group of [standard_g2p](https://github.com/tabahi/CharsiuG2P). Each 5 ms frame is classified
from at most 120 ms of audio (38.9 ms receptive field), so the model cannot learn any language's
phonotactics. Phone boundaries come from a segmental Viterbi decoder over the known phone
sequence, refined to sub-frame precision at the crossing of neighbouring posteriors.

| head | labels | size |
|---|---|---|
| `ph` | local tokens of `brahmic` (incl. `<blank> SIL noise <unk>`) | 96 |
| `phg` | gold phoneme groups, shared by every language group | 15 |
| `tone` | tone values, shared by every language group; present but untrained (no language in this group has a tone layer) | 22 |

Languages (the FLEURS languages it was trained on): hi, or, ta. standard_g2p maps every member of the group
onto the same tokens, so other members may align too, untested.

## Training

**Data:** FLEURS brahmic, all 16.8 h, training noise_level 0.02.  
**Checkpoint:** experiment `mb01a`, epoch 4, selected on FLEURS val_loss (best of 20 epochs; final epoch 3.707).  
Trunk and phg_head from latin ma02a (final epoch), ph_head and tone_head trained fresh.

| metric | value |
|---|---|
| `val_loss` | 3.5548 |
| `val_frame_acc` | 0.2727 |
| `val_frame_acc_groups` | 0.4633 |

`val_*` are on held-out FLEURS clips but are scored against the model's own alignments (there are no boundary labels outside English), so they measure self-consistency, not boundary accuracy.

Labels are standard_g2p dictionary pronunciations (gold inventory `9438371ed6dd`),
not phonetic transcriptions of what was said.

## Usage

The code is in [https://github.com/tabahi/bfa_models](https://github.com/tabahi/bfa_models) (`p4mbfa/`); clone it and run from its root.

```python
from p4mbfa.inference import MbfaAligner

aligner = MbfaAligner.from_pretrained("Tabahi/mbfa-brahmic")
wav = aligner.load_audio("clip.wav")                     # mono, 16000 Hz
phones = ["SIL", "h", "ɛ", "l", "o", "SIL"]   # this group's tokens (config.json labels.tokens)
for seg in aligner.align(wav, phones):
    print(seg["token"], seg["start_ms"], seg["end_ms"])
```

`aligner.encode(wav)` returns the per-frame log posteriors of all three heads.
Text -> tokens is standard_g2p's job (`goldG2P.phonemize_sentence` then
`lang_group_inventory.to_local`); `config.json` lists the token strings. `from_pretrained`
downloads only `config.json` and `model.safetensors`.

## Fine-tuning

`ckpt/brahmic_fleurs17h_mb01a_e4_val_loss=3.555.ckpt` is the training checkpoint (PyTorch Lightning, weights only, a pickle: load
it only if you trust this repo). Download it, then point a p4mbfa yaml at it to continue
training, or to train a new language group on its trunk and shared `phg` head:

```bash
hf download Tabahi/mbfa-brahmic "ckpt/brahmic_fleurs17h_mb01a_e4_val_loss=3.555.ckpt" --local-dir tmp/hf/mbfa-brahmic
```

```yaml
ckpt_path: "tmp/hf/mbfa-brahmic/ckpt/brahmic_fleurs17h_mb01a_e4_val_loss=3.555.ckpt"
reset_fine_heads: true      # new lang_group: rebuild ph_head / tone_head; false = same group
```

License: AGPL-3.0.