Instructions to use addisai/addis-scribe-streaming with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- NeMo
How to use addisai/addis-scribe-streaming with NeMo:
import nemo.collections.asr as nemo_asr asr_model = nemo_asr.models.ASRModel.from_pretrained("addisai/addis-scribe-streaming") transcriptions = asr_model.transcribe(["file.wav"]) - Notebooks
- Google Colab
- Kaggle
Download serving.json from addisai/addis-scribe-streaming: direct link, hf CLI and curl.
- Browser
- Download file 613 Bytes
-
https://huggingface.co/addisai/addis-scribe-streaming/resolve/main/serving.json
- Command line
-
hf download hf://addisai/addis-scribe-streaming/serving.json
-
curl -L -o serving.json https://huggingface.co/addisai/addis-scribe-streaming/resolve/main/serving.json
613 Bytes
| { | |
| "model": "addis-scribe-streaming.nemo", | |
| "architecture": "Cache-aware FastConformer (NVIDIA Nemotron 3.5 ASR Streaming 0.6B base), hybrid RNNT+CTC, greedy RNNT decoding", | |
| "sample_rate": 16000, | |
| "language_prompt": "am-ET", | |
| "language_prompt_index": 49, | |
| "default_chunk": "320ms", | |
| "chunks": { | |
| "320ms": { | |
| "att_context_size": [ | |
| 56, | |
| 3 | |
| ], | |
| "seconds": 0.32 | |
| }, | |
| "1120ms": { | |
| "att_context_size": [ | |
| 56, | |
| 13 | |
| ], | |
| "seconds": 1.12 | |
| } | |
| }, | |
| "decoding": "greedy RNNT", | |
| "output": "plain Amharic text, no punctuation", | |
| "hop_samples": 160, | |
| "n_fft": 512, | |
| "name": "Addis Scribe Streaming" | |
| } |