Stt Pl Fastconformer Hybrid Ctc Large GGUF
GGUF conversions of the CTC branch of nvidia/sttplfastconformerhybridlargepc for CrispASR. The upstream model is a hybrid transducer+CTC Polish ASR release; the shared FastConformer encoder plus the auxiliary CTC...
Modellquelle
- Art
- Other
- Dateigröße
- 66,4–209 MB
- Basismodell
- nvidia/stt_pl_fastconformer_hybrid_large_pc
- Version
- vRevision 99b0251
- Aktualisiert
- 26. Sept. 2026
- Downloads
- 352
- Quelle
- Hugging Face
Quellenbeschreibung
stt-pl-fastconformer-hybrid-ctc-large-GGUF
GGUF conversions of the CTC branch of nvidia/stt_pl_fastconformer_hybrid_large_pc for CrispASR. The upstream model is a hybrid transducer+CTC Polish ASR release; the shared FastConformer encoder plus the auxiliary CTC head are extracted here as a standalone CTC model (the RNNT prediction network and joint are dropped), giving a compact Polish ASR and forced-alignment model with punctuation + capitalisation.
| Quant | Size | Description |
|---|---|---|
| F16 | 219 MB | Full precision |
| Q8_0 | 130 MB | 8-bit |
| Q4_K | 82 MB | 4-bit K-quant (recommended) |
Architecture
17-layer NeMo FastConformer encoder + Conv1d CTC head. d_model=512, 8 heads, SentencePiece vocab, 80 log-mel features, ~115M params.
Usage
# Polish ASR:
crispasr --backend fastconformer-ctc -m stt-pl-fastconformer-hybrid-ctc-large-q4_k.gguf -f audio.wav
# Forced alignment (word timestamps for known text, or re-timing an .srt):
crispasr --align-only -am stt-pl-fastconformer-hybrid-ctc-large-q4_k.gguf \
-f audio.wav --text-file subtitles.srt --align-output retimed.srt
Attribution
All credit for the model goes to NVIDIA's NeMo team; this repository only repackages the CTC branch in GGUF form under the same CC-BY-4.0 license. Conversion: models/convert-stt-fastconformer-ctc-to-gguf.py in CrispASR.
Provenance and EU AI Act Art. 53 note
- Upstream model: nvidia/stt_pl_fastconformer_hybrid_large_pc — published by
nvidia. - Upstream licence:
cc-by-4.0. This repository redistributes under the same terms; it grants no rights the upstream licence does not. - What was done here: format conversion and/or quantisation only (GGUF). No training, no fine-tuning, no merging, no distillation, no change to architecture, vocabulary or capability. Only the numeric representation of the upstream weights differs.
- Training data: documented — where it is documented at all — by the upstream provider; see the upstream model card. No training data was used, added or selected by this repository.
- Provider status: under Regulation (EU) 2024/1689 the upstream authors remain the provider of this model. Converting the serialisation format does not make this repository the provider of a new general-purpose AI model, and no such claim is made. Questions about training content, copyright policy or model capability belong upstream.
Quellen
1 Quelle
Quellen
1 QuelleVerifiziert 6. Sept.
Modellartefakte
3 Artefakte
Modellartefakte
3 Artefaktestt-pl-fastconformer-hybrid-ctc-large-f16.gguf
gguf · 209 MB · SHA-256 696b9720e2c8…7548 · Hugging Face
Herunterladenstt-pl-fastconformer-hybrid-ctc-large-q4_k.gguf
gguf · 66,4 MB · SHA-256 4e059899cd7d…1c4b · Hugging Face
Herunterladenstt-pl-fastconformer-hybrid-ctc-large-q8_0.gguf
gguf · 112 MB · SHA-256 fa12b5fc663d…81ec · Hugging Face
HerunterladenQuellenauszüge
2 Auszüge
Quellenauszüge
2 Auszüge--- license: cc-by-4.0 base_model: nvidia/stt_pl_fastconformer_hybrid_large_pc language: - pl tags: - automatic-speech-recognition - forced-alignment - gguf - crispasr - fastconformer - ctc - nemo pipeline_tag: automatic-speech-recognition --- # stt-pl-fastconformer-hybrid-ctc-large-GGUF GGUF conversions of the **CTC branch** of [nvidia/stt_pl_fastconformer_hybrid_large_pc](https://huggingface.co/nvidia/stt_pl_fastconformer_hybrid_large_pc) for [CrispASR](https://github.com/CrispStrobe/CrispASR). The upstream model is a hybrid transducer+CTC Polish ASR release; the shared FastConformer encoder plus the auxiliary CTC head are extracted here as a standalone CTC model (the RNNT prediction network and joint are dropped), giving a compact Polish ASR **and forced-alignment** model with punctuation + capitalisation. | Quant | Size | Description | |---|---|---| | F16 | 219 MB | Full precision | | Q8_0 | 130 MB | 8-bit | | Q4_K | 82 MB | 4-bit K-quant (recommended) | ## Architecture 17-layer NeMo FastConformer encoder + Conv1d CTC head. d_model=512, 8 heads, SentencePiece vocab, 80 log-mel features, ~115M params. ## Usage ```bash # Polish ASR: crispasr --backend fastconformer-ctc -m stt-pl-fastconformer-hybrid-ctc-large-q4_k.gguf -f audio.wav # Forced alignment (word timestamps for known text, or re-timing an .srt): crispasr --align-only -am stt-pl-fastconformer-hybrid-ctc-large-q4_k.gguf \ -f audio.wav --text-file subtitles.srt --align-output retimed.srt ``` ## Attribution All credit for the model goes to NVIDIA's NeMo team; this repository only repackages the CTC branch in GGUF form under the same CC-BY-4.0 license. Conversion: `models/convert-stt-fastconformer-ctc-to-gguf.py` in CrispASR. ## Provenance and EU AI Act Art. 53 note - **Upstream model:** [nvidia/stt_pl_fastconformer_hybrid_large_pc](https://huggingface.co/nvidia/stt_pl_fastconformer_hybrid_large_pc) — published by `nvidia`. - **Upstream licence:** `cc-by-4.0`. This repository redistributes under the same terms; it grants no rights the upstream licence does not. - **What was done here:** format conversion and/or quantisation only (GGUF). No training, no fine-tuning, no merging, no distillation, no change to architecture, vocabulary or capability. Only the numeric representation of the upstream weights differs. - **Training data:** documented — where it is documented at all — by...
Source context: 352 downloads · 0 likes · Pipeline automatic-speech-recognition · Library nemo · Repo cstr/stt-pl-fastconformer-hybrid-ctc-large-GGUF
