---
license: cc-by-4.0
language:
  - th
library_name: nemo
pipeline_tag: automatic-speech-recognition
base_model: scb10x/typhoon-asr-realtime
tags:
  - automatic-speech-recognition
  - speech
  - streaming
  - FastConformer
  - Transducer
  - Thai
  - NeMo
---

# Typhoon ASR Streaming 115M

Cache-aware **streaming** Thai ASR (FastConformer-Transducer, 115M params). Converted from
[Typhoon ASR Real-time](https://huggingface.co/scb10x/typhoon-asr-realtime) — causal
convolutions + chunked limited-context attention swapped in, every weight copied as a warm
start, then briefly fine-tuned on the ~11k-hour Thai corpus. Unlike the full-context base
(which collapses to 62.8% CER when forced to stream at 1040 ms), this model decodes live,
chunk by chunk, with the encoder cache carried across chunks.


## At a glance

| | |
|---|---|
| Architecture | FastConformer-Transducer (RNN-T), cache-aware streaming |
| NeMo class | `EncDecRNNTBPEModel` |
| Tokenizer | single 2048-token Thai BPE |
| Latency knob | `chunk_size` × 80 ms → 1040 / 480 / 80 ms |
| Multi-latency `att_context_size` | `[[70,13],[70,6],[70,1],[70,0]]` |
| Streaming CER (TVSpeech) | **19.4%** @1040 ms · 16.4% @3200 ms |
| RTF (batch-1, H100 @1040 ms) | 0.020 (~50× real time) |
| Paired n-gram | [`typhoon-ai/typhoon-asr-streaming-115m-ngram`](https://huggingface.co/typhoon-ai/typhoon-asr-streaming-115m-ngram) |

Use this model for low latency / low cost when English code-switching is light; for best
accuracy use
[`typhoon-ai/typhoon-asr-streaming-nemotron-0.6b`](https://huggingface.co/typhoon-ai/typhoon-asr-streaming-nemotron-0.6b).

## Usage

Requires NVIDIA NeMo with cache-aware streaming support (see the
[project repo](https://github.com/warit-s/typhoon-asr-streaming) for the pinned commit and
the full streaming loop / server / demo):

```python
import nemo.collections.asr as nemo_asr

model = nemo_asr.models.ASRModel.restore_from("typhoon-asr-streaming-115m.nemo",
                                              map_location="cuda")
model.eval()

# offline (whole file)
text = model.transcribe(audio=["clip.wav"])[0].text

# true streaming: model.encoder.setup_streaming_params(chunk_size=13, left_chunks=2,
# shift_size=13) + conformer_stream_step(...) — see demo/asr_engine.py in the repo.
```

Decode-time vocabulary steering (phrase boosting + n-gram shallow fusion) runs inside the
streaming decoder — see the repo's `docs/SHALLOW_FUSION.md`.

## License & attribution

Released under **CC-BY-4.0**, following the base-model chain:

- fine-tuned + converted from **[scb10x/typhoon-asr-realtime](https://huggingface.co/scb10x/typhoon-asr-realtime)** (CC-BY-4.0, SCB DataX),
- itself fine-tuned from **[nvidia/stt_en_fastconformer_transducer_large](https://huggingface.co/nvidia/stt_en_fastconformer_transducer_large)** (CC-BY-4.0, NVIDIA).

Modifications in this release: full-context → cache-aware streaming conversion (causal
convolutions, chunked limited-context attention, multi-latency `att_context_size`) and a
1-epoch Thai fine-tune on the Typhoon ASR Real-time corpus.