anteju

[ASR] Add pretrained ASR models for Croatian

1

Signed-off-by: Ante Jukić # What does this PR do ? This PR adds pre-trained models for Croatian. **Collection**: [ASR] # Changelog - updated `PretrainedModelInfo` and docs - minor (cosmetic) modifications...

[Speech Data Explorer] An option to explicitly specify the base dir for manifests with relative paths **Collection**: Tools # Changelog - Use `args.audio_base_path` instead of `args.manifest` - If `args.audio_base_path` is...

# What does this PR do ? This PR aims to add a tool for generating a set of multi-channel RIRs. **Collection**: ASR # Changelog - Added `RoomCorpusGenerator` in `asr/data/data_simulation.py`...

[Fix] Use number of channels when calculating BAN

9

It seems the current implementation is missing a scaling by $M^{-1/2}$ when calculating BAN. This results in a gain of $10 \log_{10} M~\text{dB}$, which sometimes results in clipping depending on...

Audio model collection

2

# What does this PR do ? This PR moves audio-specific models to a separate collection. **Collection**: ASR, audio # Changelog - Moved audio-specific models from ASR collection into a...

[audio] Improve test coverage for audio losses

1

> [!IMPORTANT] > The `Update branch` button must only be pressed in very rare occassions. > An outdated branch is never blocking the merge of a PR. > Please reach...