Spaces:
Runtime error
Runtime error
File size: 2,605 Bytes
1e3b8aa |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 |
# Things that might be relevant
## Trained models
ESPnet model for Yoloxochitl Mixtec
- Huggingface Hub page
- Model source code https://github.com/espnet/espnet/tree/master/egs/yoloxochitl_mixtec/asr1
- Colab notebook to setup and apply the model https://colab.research.google.com/drive/1ieoW2b3ERydjaaWuhVPBP_v2QqqWsC1Q?usp=sharing
Coqui model for Yoloxochitl Mixtec
- Huggingface Hub page
- Coqui page
- Colab notebook to setup and apply the model https://colab.research.google.com/drive/1b1SujEGC_F3XhvUCuUyZK_tyUkEaFZ7D?usp=sharing#scrollTo=6IvRFke4Ckpz
Spanish ASR models
- XLS-R model based on CV8 with LM https://huggingface.co./jonatasgrosman/wav2vec2-xls-r-1b-spanish
- XLSR model based on CV6 with LM https://huggingface.co./jonatasgrosman/wav2vec2-large-xlsr-53-spanish
- XLSR model based on Librispeech https://huggingface.co./IIC/wav2vec2-spanish-multilibrispeech
Speechbrain Language identification on Common Language (from Common Voice 6/7?)
- source code https://github.com/speechbrain/speechbrain/tree/develop/recipes/CommonLanguage
- HF Hub model page https://huggingface.co./speechbrain/lang-id-commonlanguage_ecapa
- HF Hub space https://huggingface.co./spaces/akhaliq/Speechbrain-audio-classification
Speechbrain Language identification on VoxLingua
- source code https://github.com/speechbrain/speechbrain/tree/develop/recipes/VoxLingua107/lang_id
- HF Hub model page https://huggingface.co./speechbrain/lang-id-voxlingua107-ecapa
## Corpora
OpenSLR89 https://www.openslr.org/89/
Common Language https://huggingface.co./datasets/common_language
VoxLingua http://bark.phon.ioc.ee/voxlingua107/
Multilibrispeech https://huggingface.co./datasets/multilingual_librispeech
# Possible demos
## Simple categorization of utterances
A few example files are provided for each language, and the user can record their own.
The predicted confidence of each class label is shown.
## Segmentation and identification
Recordings with alternating languages in a single audio file, provided examples or the user can record.
Some voice activity detection to split the audio, then predict language of each piece
## Identication and transcription
Example files for each language separately.
The lang-id model predicts what language it is.
The corresponding ASR model produces a transcript.
## Segmentation, identification and transcription
Recordings with alternating languages in a single audio file.
Use voice activity detection to split the audio, then predict the language of each piece
Use the corresponding ASR model to produce a transcript of each piece to display. |