HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks

granite-speech-4.1-2b-nar-gguf

memoravox/granite-speech-4.1-2b-nar-gguf

kind
gguf
quant
gguf (5-bit)
family
misc
modality
automatic-speech-recognition
root model
ibm-granite/granite-speech-4.1-2b-nar
provider
memoravox
license
apache-2.0
language
en, fr, de, es, pt
context
—
library
transcribe.cpp
vRAM (GB)
—
size on disk
—
downloads
132
likes
0
created
—
last modified
2026-07-14T00:44:13+00:00 (2026)
huggingface
https://huggingface.co/memoravox/granite-speech-4.1-2b-nar-gguf
tags
transcribe.cppggufasrspeech-to-textgranitegranite-speechnarnon-autoregressivemultilingualautomatic-speech-recognitionenfrdeesptarxiv:2603.08397arxiv:2505.08699arxiv:2603.11243arxiv:2604.12398arxiv:2604.22817arxiv:2604.11269base_model:ibm-granite/granite-speech-4.1-2b-narbase_model:quantized:ibm-granite/granite-speech-4.1-2b-narlicense:apache-2.0

Benchmark scores (12)

benchmarkcategoryscorecontextconfevidence
WER Common Voicespeech-asr15.0mediumCommonVoice 15
WER Earnings-22speech-asr8.48Earnings-22medium| Earnings-22 | 8.48 | GigaSpeech | 10.12 |
WER Common Voicespeech-asr7.17CommonVoice 15 FRmedium| CommonVoice 15 FR | 7.17 | AMI IHM | 7.91 |
WER Common Voicespeech-asr6.5CommonVoice 15 ENmedium| CommonVoice 15 EN | 6.50 | MLS ES | 3.31 |
WER VoxPopulispeech-asr5.83VoxPopulimedium| VoxPopuli | 5.83 | | |
WER Common Voicespeech-asr4.73CommonVoice 15 DEmedium| CommonVoice 15 DE | 4.73 | MLS FR | 4.52 |
WER Common Voicespeech-asr4.02CommonVoice 15 ESmedium| CommonVoice 15 ES | 4.02 | MLS PT | 11.86 |
WER SPGISpeechspeech-asr3.04SPGISpeechmedium| SPGISpeech | 3.04 | TED-LIUM | 3.67 |
WER LibriSpeechspeech-asr2.75LibriSpeech othermedium| LibriSpeech other | 2.75 | MLS DE | 4.75 |
WER Common Voicespeech-asr2.57CommonVoice 15 PTmedium| CommonVoice 15 PT | 2.57 | AMI SDM | 19.59 |
WERspeech-asr1.29low| BF16 | [granite-speech-4.1-2b-nar-BF16.gguf](https://huggingface.co/handy-computer/granite-speech-4.1-2b-nar-gguf/resolve/main/granite-speech-4.1-2b-nar-BF16.gguf) | 4.20 GB | 1.
WER LibriSpeechspeech-asr1.29LibriSpeech cleanmedium| LibriSpeech clean | 1.29 | MLS EN | 4.77 |

Discussions (0)

none scraped.

Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.