diff --git a/src-tauri/Cargo.toml b/src-tauri/Cargo.toml
index 46773382a..4e317c4cb 100644
--- a/src-tauri/Cargo.toml
+++ b/src-tauri/Cargo.toml
@@ -105,7 +105,7 @@ transcribe-rs = { version = "0.3.3", features = ["whisper-metal"] }
 [target.'cfg(target_os = "linux")'.dependencies]
 gtk-layer-shell = { version = "0.8", features = ["v0_6"] }
 gtk = "0.18"
-transcribe-rs = { version = "0.3.3", features = ["whisper-vulkan"] }
+transcribe-rs = { version = "0.3.8", features = ["whisper-cpp", "onnx", "whisper-vulkan", "ort-rocm", "ort-cuda"] }
 
 [patch.crates-io]
 tauri-runtime = { git = "https://github.com/cjpais/tauri.git", branch = "handy-2.10.2" }
diff --git a/src/i18n/locales/en/translation.json b/src/i18n/locales/en/translation.json
index eef96e5be..10e8dca0f 100644
--- a/src/i18n/locales/en/translation.json
+++ b/src/i18n/locales/en/translation.json
@@ -258,11 +258,11 @@
       "acceleration": {
         "whisper": {
           "title": "Whisper Acceleration",
-          "description": "Hardware acceleration for Whisper models. Auto uses GPU if available (Metal on macOS, Vulkan on Windows/Linux)."
+          "description": "GPU acceleration for Whisper models only. Does not affect Parakeet, Canary, or Moonshine — those use ONNX Acceleration below. Auto uses Metal on macOS, Vulkan on Windows/Linux."
         },
         "ort": {
           "title": "ONNX Acceleration",
-          "description": "Hardware acceleration for ONNX models (Parakeet, Canary, Moonshine, etc.). DirectML on Windows is experimental. Models may fail to transcribe."
+          "description": "GPU acceleration for ONNX models: Parakeet, Canary, Moonshine, SenseVoice, etc. Does not affect Whisper models. Linux: ROCm (AMD) or CUDA (NVIDIA). Windows: DirectML. macOS: CPU only."
         },
         "gpuDevice": {
           "title": "GPU Device",