Update README.md
Browse files
README.md
CHANGED
|
@@ -8,65 +8,22 @@ pinned: false
|
|
| 8 |
---
|
| 9 |
|
| 10 |
# NTNU × eVici Voice Ordering
|
|
|
|
| 11 |
|
| 12 |
-
|
|
|
|
| 13 |
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
Farmers often need to place animal feed orders while working in environments where:
|
| 21 |
-
|
| 22 |
-
- hands are occupied
|
| 23 |
-
- gloves or machinery make typing difficult
|
| 24 |
-
- screens are inconvenient or unsafe to use
|
| 25 |
-
|
| 26 |
-
Traditional web or mobile ordering systems rely on **manual, form-based interaction**, which is slow and impractical in these situations.
|
| 27 |
-
|
| 28 |
-
We propose a **voice-driven ordering module**:
|
| 29 |
-
|
| 30 |
-
> A speech-controlled interface that enables hands-free, natural-language order placement directly integrated into eVici’s existing ordering solution.
|
| 31 |
-
|
| 32 |
-
The system leverages:
|
| 33 |
-
|
| 34 |
-
- domain-specific language understanding (NLU/SLU)
|
| 35 |
-
- intent detection
|
| 36 |
-
- slot filling / entity extraction
|
| 37 |
-
- fuzzy matching of products and locations
|
| 38 |
-
- dialogue management for clarification and robustness
|
| 39 |
-
- low-latency inference suitable for real-world deployment
|
| 40 |
-
|
| 41 |
-
The aim is **accurate, robust, and efficient order creation in real agricultural environments**.
|
| 42 |
-
|
| 43 |
-
---
|
| 44 |
-
|
| 45 |
-
## What we publish here
|
| 46 |
-
|
| 47 |
-
This organization may include:
|
| 48 |
-
|
| 49 |
-
- 📦 Synthetic and annotated Norwegian SLU datasets
|
| 50 |
-
- 🤖 Fine-tuned Transformer/BERT models
|
| 51 |
-
- 🧪 Experimental baselines and benchmarks
|
| 52 |
-
- 🗣️ Speech/NLU prototypes
|
| 53 |
-
- 📊 Evaluation tools and scripts
|
| 54 |
-
|
| 55 |
-
All resources support research and prototyping of **voice-based industrial ordering systems**.
|
| 56 |
-
|
| 57 |
-
---
|
| 58 |
|
| 59 |
## Scope
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
- domain-specific (animal feed ordering)
|
| 64 |
-
- small-data and applied
|
| 65 |
-
- Norwegian-language focused
|
| 66 |
-
- research and prototype quality
|
| 67 |
-
|
| 68 |
-
They are not general-purpose language models.
|
| 69 |
|
| 70 |
---
|
| 71 |
|
| 72 |
-
*Developed as part of a computer
|
|
|
|
| 8 |
---
|
| 9 |
|
| 10 |
# NTNU × eVici Voice Ordering
|
| 11 |
+
Bachelor’s thesis at NTNU with eVici AS – voice‑driven animal‑feed ordering for farmers.
|
| 12 |
|
| 13 |
+
## Vision
|
| 14 |
+
Hands‑free, natural‑language ordering that works while farmers’ hands are busy, gloved, or near machinery.
|
| 15 |
|
| 16 |
+
## We publish
|
| 17 |
+
- Synthetic & annotated Norwegian SLU datasets
|
| 18 |
+
- Fine‑tuned Transformer/BERT models
|
| 19 |
+
- Baselines, benchmarks & evaluation scripts
|
| 20 |
+
- Speech/NLU prototypes
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
## Scope
|
| 23 |
+
- Domain‑specific (feed ordering)
|
| 24 |
+
- Small‑data, Norwegian‑focused
|
| 25 |
+
- Research‑prototype quality (not a general‑purpose LM)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
|
| 27 |
---
|
| 28 |
|
| 29 |
+
*Developed as part of a computer‑science bachelor thesis at NTNU in partnership with eVici AS.*
|