TTSModel — Swahili TTS
A Swahili text-to-speech model, finetuned from Meta's MMS-TTS Swahili checkpoint on the swa_tts split of google/WaxalNLP, using the VITS finetuning recipe from ylacombe/finetune-hf-vits. Developed by the Maseno Centre for Applied AI (MCAAI).
Model details
- Base model:
facebook/mms-tts-swh(Meta's Massively Multilingual Speech TTS, Swahili) - Architecture: VITS (single-speaker)
- Training data: google/WaxalNLP,
swa_ttsconfig — 1,387 train utterances (805 after duration/length filtering), single Swahili speaker, 16kHz audio, sourced via the Loud & Clear initiative. - Training checkpoint: step 4,500 / 20,200 planned steps
- Sample rate: 16,000 Hz
- Language: Swahili (
swh/ ISO 639-3)
Training configuration
| Setting | Value |
|---|---|
| Learning rate | 2e-5 |
| Batch size | 8 |
| Precision | fp16 |
| Max clip duration | 20s |
| Min clip duration | 0.5s |
| Loss weights | mel=35, kl=1.5, disc=3, gen/fmaps/duration=1 |
Training used the finetune-hf-vits recipe with transformers==4.35.1, datasets==2.14.7, accelerate==0.24.1, numpy<2.0.
Usage
import numpy as np
from transformers import pipeline
from IPython.display import Audio as IPyAudio
synthesiser = pipeline("text-to-speech", model="MCAA1-MSU/TTSModel")
speech = synthesiser("Habari yako, karibu Kenya.")
audio = np.squeeze(speech["audio"])
display(IPyAudio(audio, rate=speech["sampling_rate"]))
Verified loading cleanly with transformers pipeline("text-to-speech", ...) — no missing/unexpected weight warnings on load.
Intended use
Research and experimentation with Swahili TTS.
Limitations
- Single speaker, single dataset: may not generalize well to varied Swahili dialects, accents, or speaking styles.
- No formal evaluation yet: no MOS, WER, or MCD scores computed. Qualitative spot-checks suggest intelligible but rough output.
- Small dataset: ~800 training utterances after filtering.
License
Derived from facebook/mms-tts-swh (CC-BY-NC-4.0, non-commercial). This finetuned model inherits that license. swa_tts training data is CC-BY-SA-4.0.
Acknowledgements
- Meta MMS for the base checkpoint
- WAXAL / google/WaxalNLP for the training data
- ylacombe/finetune-hf-vits for the finetuning recipe