Xuesong Yang

I am using this backend to make a G2P-like conversation. But I did not know where to find a complete list of the phone set for each language. Could anyone...

Signed-off-by: Xuesong Yang # What does this PR do ? Add a one line overview of what this PR aims to accomplish. **Collection**: [Note which collection this PR will affect]...

# What does this PR do ? **Collection**: [Note which collection this PR will affect] # Changelog - Add specific line by line info of high level changes in this...

broken link for tts tutorials.

1

The link (https://docs.nvidia.com/deeplearning/riva/user-guide/docs/tutorials/tts-python-basics-and-customization-with-ssml.html) shared in https://docs.nvidia.com/deeplearning/riva/user-guide/docs/tutorials/tts-basics-customize-ssml.html#note is not valid anymore.

[zh] WARNING:NeMo-text-processing:Failed text: 免除GOOGLE在一桩诽谤官司中的法律责任。Key: integer_part Value: None

8

Received warning message when normalizing text. Could you pls provide what the message indicates? **Reproduciple code**: ```python from nemo_text_processing.text_normalization.normalize import Normalizer text_normalizer = Normalizer(lang="zh", input_case="cased", overwrite_cache=True, cache_dir=str("cache_dir")) text_normalizer_call_kwargs = {"punct_pre_process":...

maybe wrong IPA symbol for a German word "aneinander"

13

Previously reported to Phonemizer repo: https://github.com/bootphon/phonemizer/issues/142, but suggested to re-direct here. There is `/1/` appearing in the phonetic transcription for the German language. Is it expected? ``` $ echo "aneinander"...

* tensorboard and wandb save hparam config files with slightly different YAML structure. This fix will address blockers when feeding in `*.ckpt` and `*/wandb/latest-run/files/config.yaml`. * simplified loading ASR models by...

- change the codes names in cuts into target_codes and context_codes. - remove `codec_model_name` in yaml config because we've added the codec name in `input_cfg.yaml` file. - fixed bugs that...

[magpietts][lhotse_v2] Adding scripts of converting NeMo manifests to Lhotse Shars, and speedup improvements for codec model inference.

9

### Speed Up Codec Model Inference Made slightly changes for the dependency util func `mask_sequence_tensor`. We observed at least 3x speedup. This change is cherry-picked from the part of PR...

[magpietts] added logging of consumed samples at each training step.

1

The Lhotse dataloader applies dynamic batching based on duration buckets, causing the number of consumed samples at each training step to vary. Tracking the number of consumed samples enables fairer...