File size: 816 Bytes
f85841b d53231b f85841b d807327 43d19dc 3d94da9 d53231b 3d94da9 d53231b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 |
---
license: mit
language:
- en
- de
- da
- 'no'
- sv
pipeline_tag: automatic-speech-recognition
---
# Stremma ELM(extendable language model)
The initial model that was used for https://stremma.ai service, serves as a baseline for the current v2 in production, handling most of the ASR tasks.
## Detais
Stremma ELM is a transformer-based encoder-decoder model, also referred to as a sequence-to-sequence model. It was trained on 0.1M hours of labeled audio and 0.4M hours of pseudolabeled audio collected via Stremma SaaS during period of work in manual transcribing.
The models were trained on either English or Danish data. The English and Danish models were trained on the task of speech recognition. The multilingual extensions were trained on both speech recognition and speech translation.
## Usage
TBD |