Capability matrix
- Compute and format Mixed FP16/FP32 · Historical Python reference on Ryzen 9 9950X3D: five-minute RTF 0.220–0.225, about 8.51 GB peak RSS.
- Local entry point Audio transcription.
- Server Unverified for this artifact. Check for a service that accepts this artifact.
- Harness No recorded client task. Run one identified request, then a task with a checked result.
Choose the download
Publisher model card · CKE input files
Publisher task: automatic speech recognition. CKE example: GGUF, mixed FP16/FP32. Use this artifact identifier:
hf://cstr/cohere-transcribe-03-2026-GGUF/cohere-transcribe.gguf.
Get started
Generated-C frontend/encoder and decoder components connected by a native transcription host; externally supplied segments for long audio. See the Cohere kernels and Transcribe guide. Use the model-specific setup.
What has actually run?
Generated-C frontend/encoder and decoder libraries under a native C host; historical Python reference: exact short trajectory and complete 42-minute VAD-sliced run. English audio with externally supplied speech segments; native VAD, independent timestamp accuracy, multilingual and concurrent-session certification remain open.
CPU evidence. Historical Python reference on Ryzen 9 9950X3D: five-minute RTF 0.220–0.225, about 8.51 GB peak RSS; 42-minute RTF 0.252, 8.75 GB peak RSS. Not generated-runtime benchmarks..
Performance. For speech, compare transcription time with audio duration and check the transcript against known words. ARM throughput for this exact artifact is unverified here. Benchmark evidence · 32K to 128K results
Can I fine-tune it?
Full-checkpoint training, LoRA and other PEFT methods: no verified CKE recipe for this exact checkpoint in the reviewed evidence. For a trainable starter, use the FP32 and adapter recipes.
Transcribe with prepared encoder and decoder bundles
Once you have the encoder and decoder bundles, substitute their paths below. For a long recording, prepare speech segments separately.
make ck-audio-encoder-decoder-transcribe-v8
./build/ck-audio-encoder-decoder-transcribe-v8 \
/path/to/cohere-encoder/libmodel.so \
/path/to/cohere-encoder/weights.bump \
/path/to/cohere-encoder/weights_manifest.map \
/path/to/cohere-decoder/libmodel.so \
/path/to/cohere-decoder/weights.bump \
/path/to/cohere-decoder/weights_manifest.map \
recording.wav segments.txt