Downloads · 30 days
19
20% of all-time downloads
mms-1b-all-qxp
robertp408/mms-1b-all-qxp
mms-1b-all-qxp is a automatic speech recognition model from robertp408. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as apache-2.0.
This model was created from the Mozilla.ai Blueprint: speech-to-text-finetune.
All-time downloads
94
Public
Parameters
965M
38.6 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors3.9 GB · 100%
At a glance
- Task
- Automatic Speech Recognition
- Library
- transformers
- License
- apache-2.0
- Model type
- wav2vec2
- Access
- Public
- Created
- Nov 19, 2025
- Updated
- Nov 20, 2025
- SHA
- bb0f4e49
Base models
Datasets
Model card
From the Hugging Face model README
Finetuned facebook/mms-1b-all on 1502 Puno Quechua training audio samples from cmflnuzw781hfx067jwv4pt9x.
This model was created from the Mozilla.ai Blueprint: speech-to-text-finetune.
Evaluation results on 562 audio samples of Puno Quechua:
Baseline model (before finetuning) on Puno Quechua
- Word Error Rate (Normalized): 146.227
- Word Error Rate (Orthographic): 100.0
- Character Error Rate (Normalized): 332.297
- Character Error Rate (Orthographic): 344.813
- Loss: 23.528
Finetuned model (after finetuning) on Puno Quechua
- Word Error Rate (Normalized): 19.5
- Word Error Rate (Orthographic): 32.432
- Character Error Rate (Normalized): 2.718
- Character Error Rate (Orthographic): 4.452
- Loss: 0.163
- Task
- Automatic Speech Recognition
- Library
- transformers
- Type
- wav2vec2
- License
- apache-2.0
- Languages
- qxp
- Created
- Nov 19, 2025
- Updated
- Nov 20, 2025