CS Thesis May 2026
Collection
Adja MT, ASR, TTS datasets and models for Josue Godeme's May 2026 CS thesis. GitHub: github.com/FrejusGdm/cs-thesis-may-2026 • 9 items • Updated
Datasets:
Dataset Viewer
The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
CS Thesis May 2026 ASR Results
This dataset repo is the canonical public ASR result archive for the May 2026 Adja thesis release. It contains reports, metrics, and selected predictions only. It intentionally does not contain checkpoints, raw private audio, or active experiment dependency folders.
Contents
reports/: ASR experiment reports, conclusions, progress summaries, and pipeline notes.reports/experiments/: per-experiment report folders for C/E tracks.predictions/: selected public decode samples and qualitative ASR outputs.manifests/: export manifest with source paths and release revision.
Best Public ASR Artifacts
- Primary ASR model:
JosueG/wav2vec2-xlsr-adja-c4v2 - Complementary ASR model:
JosueG/whisper-ewe-adja-e4v4 - Best measured ASR result:
Omni_AUDIOFIX2_LLM7B_20260509_M0, test CER19.86, test WER62.28, exported underreports/omni_audiofix2_20260509/ - Canonical speech dataset:
JosueG/adja-speech-asr-tts - GitHub release:
FrejusGdm/cs-thesis-may-2026
Provenance
The old mixed JosueG/adja-asr-results repo remains provenance for checkpoints
and active experiments. This export is the public reporting layer.
- Downloads last month
- 67