Migrate to transcribe-rs-0.3.1 and add Canary support#1023
Migrate to transcribe-rs-0.3.1 and add Canary support#1023cjpais merged 4 commits intocjpais:mainfrom
Conversation
intech
force-pushed
the
feat/transcribe-rs-v0.3.0
branch
2 times, most recently
from
f81bdcf to
41ccb7b
Compare
|
@intech if you don't mind 0.3.1 is out, you can use that version here. I think we can also add Canary support directly. We need to make sure to get the languages correct as per each model card. Note these are int8 I've uploaded the models: I will be doing a once over the whole PR as well |
|
Updated to transcribe-rs v0.3.1 as requested. Here's what changed: Dependency: Canary models added:
Language verification — checked against HuggingFace model cards:
Tested locally:
i18n translations added for all 17 locales. |
Breaking API migration: - engines::* → onnx::* module paths, whisper_cpp::* for Whisper - TranscriptionEngine trait → SpeechModel trait - Two-step new()+load_model() → one-step Model::load() - transcribe_samples(Vec<f32>) → transcribe(&[f32]) - Explicit unload_model() → RAII drop - Canary engine now from transcribe-rs instead of local crate Removes canary-engine local crate dependency. Uses path dependency to local transcribe-rs clone (temporary). Co-Authored-By: Claude Opus 4.6 <[email protected]>
Update transcribe-rs dependency from v0.2.8 to v0.3.1 with restructured features (whisper + onnx). Add Canary 180M Flash (4 languages, 146MB) and Canary 1B v2 (25 EU languages, 692MB) with translation support. Add i18n translations for all 17 locales. Languages verified against HuggingFace model cards: - Canary 180M Flash: en, de, es, fr - Canary 1B v2: bg, hr, cs, da, nl, en, et, fi, fr, de, el, hu, it, lv, lt, mt, pl, pt, ro, sk, sl, es, sv, ru, uk Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
intech
force-pushed
the
feat/transcribe-rs-v0.3.0
branch
from
8ae7945 to
65b138e
Compare
🧪 Test Build ReadyBuild artifacts for PR #1023 are available for testing. Download artifacts from workflow run Artifacts expire after 30 days. |
cjpais
changed the title
2 participants
Summary
Closed #1022
Migrates Handy from
transcribe-rs v0.2.8tov0.3.0API. Canary 1B v2 is now part of transcribe-rs with fullSpeechModeltrait support.Requires: cjpais/transcribe-rs#55
Canary 1B v2 model
Model files can be downloaded from https://huggingface.co/istupakov/canary-1b-v2-onnx
Changes
src-tauri/src/managers/transcription.rs(-96/+111 lines)Imports:
transcribe_rs::engines::*totranscribe_rs::onnx::*/whisper_cpp::*,TranscriptionEnginetoSpeechModelModel loading (all 7 engines):
Engine::new()+load_model()toModel::load(path, &Quantization)Transcription:
transcribe_samples(audio, params)totranscribe(&audio, &options)/transcribe_with(&audio, ¶ms)Unloading: explicit
unload_model()to RAII dropTest plan
cargo checkcompiles