Automatic Speech Recognition
NeMo
Safetensors
PyTorch
Transformers
English
parakeet_rnnt
feature-extraction
speech
audio
Transducer
FastConformer
Conformer
NeMo
hf-asr-leaderboard
Eval Results (legacy)
Eval Results
Instructions to use nvidia/parakeet-rnnt-0.6b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- NeMo
How to use nvidia/parakeet-rnnt-0.6b with NeMo:
import nemo.collections.asr as nemo_asr asr_model = nemo_asr.models.ASRModel.from_pretrained("nvidia/parakeet-rnnt-0.6b") transcriptions = asr_model.transcribe(["file.wav"]) - Transformers
How to use nvidia/parakeet-rnnt-0.6b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="nvidia/parakeet-rnnt-0.6b")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("nvidia/parakeet-rnnt-0.6b", dtype="auto") - Notebooks
- Google Colab
- Kaggle
Update README: available in Transformers 5.13.0
#5
by eustlb HF Staff - opened
README.md
CHANGED
|
@@ -240,10 +240,10 @@ python [NEMO_GIT_FOLDER]/examples/asr/transcribe_speech.py
|
|
| 240 |
### 2) [Transformers](https://github.com/huggingface/transformers) 🤗 usage
|
| 241 |
|
| 242 |
|
| 243 |
-
|
| 244 |
|
| 245 |
```bash
|
| 246 |
-
pip install
|
| 247 |
```
|
| 248 |
|
| 249 |
<details>
|
|
@@ -252,7 +252,7 @@ pip install git+https://github.com/huggingface/transformers
|
|
| 252 |
```python
|
| 253 |
from transformers import pipeline
|
| 254 |
|
| 255 |
-
pipe = pipeline("automatic-speech-recognition", model="
|
| 256 |
out = pipe("https://huggingface.co/datasets/hf-internal-testing/dummy-audio-samples/resolve/main/bcn_weather.mp3")
|
| 257 |
print(out)
|
| 258 |
```
|
|
@@ -267,7 +267,7 @@ from datasets import load_dataset, Audio
|
|
| 267 |
|
| 268 |
num_samples = 3
|
| 269 |
|
| 270 |
-
model_id = "
|
| 271 |
processor = AutoProcessor.from_pretrained(model_id)
|
| 272 |
model = AutoModelForRNNT.from_pretrained(model_id, dtype="auto", device_map="auto")
|
| 273 |
|
|
@@ -292,7 +292,7 @@ from transformers import AutoModelForRNNT, AutoProcessor
|
|
| 292 |
|
| 293 |
num_samples = 3
|
| 294 |
|
| 295 |
-
model_id = "
|
| 296 |
processor = AutoProcessor.from_pretrained(model_id)
|
| 297 |
model = AutoModelForRNNT.from_pretrained(model_id, dtype="auto", device_map="auto")
|
| 298 |
|
|
@@ -321,7 +321,7 @@ from transformers import AutoModelForRNNT, AutoProcessor
|
|
| 321 |
from datasets import load_dataset, Audio
|
| 322 |
import torch
|
| 323 |
|
| 324 |
-
model_id = "
|
| 325 |
NUM_SAMPLES = 4
|
| 326 |
|
| 327 |
processor = AutoProcessor.from_pretrained(model_id)
|
|
|
|
| 240 |
### 2) [Transformers](https://github.com/huggingface/transformers) 🤗 usage
|
| 241 |
|
| 242 |
|
| 243 |
+
Parakeet RNNT is available in 🤗 Transformers starting from v5.13.0.
|
| 244 |
|
| 245 |
```bash
|
| 246 |
+
pip install "transformers>=5.13.0"
|
| 247 |
```
|
| 248 |
|
| 249 |
<details>
|
|
|
|
| 252 |
```python
|
| 253 |
from transformers import pipeline
|
| 254 |
|
| 255 |
+
pipe = pipeline("automatic-speech-recognition", model="nvidia/parakeet-rnnt-0.6b")
|
| 256 |
out = pipe("https://huggingface.co/datasets/hf-internal-testing/dummy-audio-samples/resolve/main/bcn_weather.mp3")
|
| 257 |
print(out)
|
| 258 |
```
|
|
|
|
| 267 |
|
| 268 |
num_samples = 3
|
| 269 |
|
| 270 |
+
model_id = "nvidia/parakeet-rnnt-0.6b"
|
| 271 |
processor = AutoProcessor.from_pretrained(model_id)
|
| 272 |
model = AutoModelForRNNT.from_pretrained(model_id, dtype="auto", device_map="auto")
|
| 273 |
|
|
|
|
| 292 |
|
| 293 |
num_samples = 3
|
| 294 |
|
| 295 |
+
model_id = "nvidia/parakeet-rnnt-0.6b"
|
| 296 |
processor = AutoProcessor.from_pretrained(model_id)
|
| 297 |
model = AutoModelForRNNT.from_pretrained(model_id, dtype="auto", device_map="auto")
|
| 298 |
|
|
|
|
| 321 |
from datasets import load_dataset, Audio
|
| 322 |
import torch
|
| 323 |
|
| 324 |
+
model_id = "nvidia/parakeet-rnnt-0.6b"
|
| 325 |
NUM_SAMPLES = 4
|
| 326 |
|
| 327 |
processor = AutoProcessor.from_pretrained(model_id)
|