Add hebtts and lothm scripts for TTS model inference setup
thewh1teagle
Unverified signature
687792c
Update axis labels in WER plot for clarity and improved readability
thewh1teagle
Unverified signature
180f090
Update WER plot with improved model metrics and visualization enhancements
thewh1teagle
Unverified signature
475a735
add all metrics
thewh1teagle
Unverified signature
6c7f537
Add text normalization and update report structure in transcription process
thewh1teagle
Unverified signature
bc8ad9c
Add rtf.json with initial performance metrics for models
thewh1teagle
Unverified signature
429b270
Add unvocalized sentences CSV and update model references in scripts
thewh1teagle
Unverified signature
ab52a44
add lothm
thewh1teagle
Unverified signature
d6383e7
add hebtts samples
thewh1teagle
Unverified signature
c812a22
update wavs
thewh1teagle
Unverified signature
4b11b1a
Update file paths for CSV and model in Piper TTS script
thewh1teagle
Unverified signature
74029f0
- Changed CSV path to point to the new dataset location:
/Users/user/Documents/audio/phonikud-experiments/comparison/sentences/saspeech_male_phonikud.csv
- Updated model path to use the new ONNX model:
/Users/user/Documents/audio/phonikud-experiments/comparison/model (4).onnx
add saspeech samples
thewh1teagle
Unverified signature
b7715c8
add stts2, sort tables
thewh1teagle
Unverified signature
26bbc8b
Update .gitignore, add piper.py for audio generation, and enhance todo list
thewh1teagle
Unverified signature
fdd064f
add wav files
thewh1teagle
Unverified signature
949d638
add wav files
thewh1teagle
Unverified signature
4fbbe4a
add mms files
thewh1teagle
Unverified signature
eb586bd
Add WER transcription script with audio processing and metrics calculation
thewh1teagle
Unverified signature
17d4b8f
- Implemented a script to transcribe audio files using a Whisper model.
- Added functions for downloading the model, preparing audio, and transcribing.
- Integrated WER and CER metrics for evaluating transcription accuracy.
- Generated JSON reports for each audio folder containing mean and individual metrics.
- Utilized pandas for data handling and tqdm for progress tracking.
Add audio files for ablation and update comparison tasks in todo.txt
thewh1teagle
Unverified signature
0ccf643
Add vocalization processing script and initial data files
thewh1teagle
Unverified signature
cbf5200
- Created `create_vocalized.py` to apply diacritics using Dicta, Phonikud, and Nakdimon models.
- Added `study_text.csv` containing unvocalized Hebrew sentences.
- Generated `study_dicta.csv`, `study_nakdimon.csv`, and `study_phonikud.csv` with vocalized outputs from respective models.
Add new datasets and audio comparison tasks
thewh1teagle
Unverified signature
29f431f
- Created a new CSV file `saspeech_nakdimon.csv` containing 100 sentences for evaluation.
- Added a `todo.txt` file outlining tasks for user study and audio comparisons.
- Introduced a new CSV file `sentences.csv` for user study with 10 example sentences.
