feat: add GigaAM v3 for Russian speech recognition (#913)

* feat: add GigaAM v3 model for Russian speech recognition Add GigaAM v3 e2e_ctc as a new transcription engine using transcribe-rs 0.2.7 gigaam feature. Russian speech recognition with punctuation, Latin characters and digit support. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: cargo fmt formatting Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Keep the file name of the model download the same as the file on the blob website. --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: CJ Pais <cj@cjpais.com>
2026-03-01 14:53:48 +01:00 · 2026-03-01 14:53:48 +01:00 · ff8612297b
commit ff8612297b
parent 17d34a910f
22 changed files with 154 additions and 13 deletions
--- a/src-tauri/Cargo.lock
+++ b/src-tauri/Cargo.lock
@ -14,6 +14,16 @@ version = "2.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "320119579fcad9c21884f5c4861d16174d0e06250625266f50fe6898340abefa"

+[[package]]
+name = "advapi32-sys"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e06588080cb19d0acb6739808aafa5f26bfb2ca015b2b6370028b44cf7cb8a9a"
+dependencies = [
+ "winapi 0.2.8",
+ "winapi-build",
+]
+
 [[package]]
 name = "ahash"
 version = "0.7.8"
@ -1392,7 +1402,7 @@ checksum = "1b1d1d91c932ef41c0f2663aa8b0ca0342d444d842c06914aa0a7e352d0bada6"
 dependencies = [
 "libc",
 "redox_users 0.4.6",
- "winapi",
+ "winapi 0.3.9",
 ]

 [[package]]
@ -1454,7 +1464,7 @@ dependencies = [
 "dlopen2_derive",
 "libc",
 "once_cell",
- "winapi",
+ "winapi 0.3.9",
 ]

 [[package]]
@ -2249,7 +2259,7 @@ dependencies = [
 "gobject-sys",
 "libc",
 "system-deps 6.2.2",
- "winapi",
+ "winapi 0.3.9",
 ]

 [[package]]
@ -3189,7 +3199,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b67380fd3b2fbe7527a606e18729d21c6f3951633d0500574c4dc22d2d638b9f"
 dependencies = [
 "cfg-if",
- "winapi",
+ "winapi 0.3.9",
 ]

 [[package]]
@ -4853,7 +4863,7 @@ dependencies = [
 "strum",
 "strum_macros",
 "widestring",
- "winapi",
+ "winapi 0.3.9",
 "x11",
 ]

@ -4872,7 +4882,7 @@ dependencies = [
 "inotify 0.8.3",
 "lazy_static",
 "libc",
- "winapi",
+ "winapi 0.3.9",
 "x11",
 ]

@ -5757,7 +5767,7 @@ checksum = "f0c3dbbd9ae980613c6dd8e28a9407b50509d3803b57624d5dfe8315218cd58b"
 dependencies = [
 "byteorder",
 "libc",
- "winapi",
+ "winapi 0.3.9",
 ]

 [[package]]
@ -7143,10 +7153,11 @@ dependencies = [

 [[package]]
 name = "transcribe-rs"
-version = "0.2.5"
+version = "0.2.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "23e3bf6e1dd4909268796bdf3794fb6fda70f4a0e5a80e2e56cd3fc213a11586"
+checksum = "25ad0a55f1482ec523073073a3a08a457c57dd956d09d62cdfe05a27a5591334"
 dependencies = [
+ "advapi32-sys",
 "base64 0.22.1",
 "derive_builder",
 "env_logger",
@ -7232,7 +7243,7 @@ checksum = "89daebc3e6fd160ac4aa9fc8b3bf71e1f74fbf92367ae71fb83a037e8bf164b9"
 dependencies = [
 "memoffset",
 "tempfile",
- "winapi",
+ "winapi 0.3.9",
 ]

 [[package]]
@ -7882,6 +7893,12 @@ version = "1.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "72069c3113ab32ab29e5584db3c6ec55d416895e60715417b5b883a357c3e471"

+[[package]]
+name = "winapi"
+version = "0.2.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "167dc9d6949a9b857f3451275e911c3f44255842c1f7a76f33c55103a909087a"
+
 [[package]]
 name = "winapi"
 version = "0.3.9"
@ -7892,6 +7909,12 @@ dependencies = [
 "winapi-x86_64-pc-windows-gnu",
 ]

+[[package]]
+name = "winapi-build"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2d315eee3b34aca4797b2da6b13ed88266e6d612562a0c46390af8299fc699bc"
+
 [[package]]
 name = "winapi-i686-pc-windows-gnu"
 version = "0.4.0"
@ -8497,7 +8520,7 @@ version = "0.10.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "80d0f4e272c85def139476380b12f9ac60926689dd2e01d4923222f40580869d"
 dependencies = [
- "winapi",
+ "winapi 0.3.9",
 ]

 [[package]]
--- a/src-tauri/Cargo.toml
+++ b/src-tauri/Cargo.toml
@ -68,7 +68,7 @@ chrono = "0.4"
 rusqlite = { version = "0.37", features = ["bundled"] }
 tar = "0.4.44"
 flate2 = "1.0"
-transcribe-rs = { version = "0.2.5", features = ["whisper", "parakeet", "moonshine", "sense_voice"] }
+transcribe-rs = { version = "0.2.7", features = ["whisper", "parakeet", "moonshine", "sense_voice", "gigaam"] }
 handy-keys = "0.2.1"
 ferrous-opencc = "0.2.3"
 clap = { version = "4", features = ["derive"] }
--- a/src-tauri/src/managers/model.rs
+++ b/src-tauri/src/managers/model.rs
@ -23,6 +23,7 @@ pub enum EngineType {
    Moonshine,
    MoonshineStreaming,
    SenseVoice,
+    GigaAM,
 }

 #[derive(Debug, Clone, Serialize, Deserialize, Type)]
@ -397,6 +398,32 @@ impl ModelManager {
            },
        );

+        // GigaAM v3 supported languages
+        let gigaam_languages: Vec<String> = vec!["ru"].into_iter().map(String::from).collect();
+
+        available_models.insert(
+            "gigaam-v3-e2e-ctc".to_string(),
+            ModelInfo {
+                id: "gigaam-v3-e2e-ctc".to_string(),
+                name: "GigaAM v3".to_string(),
+                description: "Russian speech recognition. Fast and accurate.".to_string(),
+                filename: "giga-am-v3.int8.onnx".to_string(),
+                url: Some("https://blob.handy.computer/giga-am-v3.int8.onnx".to_string()),
+                size_mb: 225,
+                is_downloaded: false,
+                is_downloading: false,
+                partial_size: 0,
+                is_directory: false,
+                engine_type: EngineType::GigaAM,
+                accuracy_score: 0.85,
+                speed_score: 0.75,
+                supports_translation: false,
+                is_recommended: false,
+                supported_languages: gigaam_languages,
+                is_custom: false,
+            },
+        );
+
        // Auto-discover custom Whisper models (.bin files) in the models directory
        if let Err(e) = Self::discover_custom_whisper_models(&models_dir, &mut available_models) {
            warn!("Failed to discover custom models: {}", e);
--- a/src-tauri/src/managers/transcription.rs
+++ b/src-tauri/src/managers/transcription.rs
@ -12,6 +12,7 @@ use std::time::{Duration, SystemTime};
 use tauri::{AppHandle, Emitter};
 use transcribe_rs::{
    engines::{
+        gigaam::GigaAMEngine,
        moonshine::{
            ModelVariant, MoonshineEngine, MoonshineModelParams, MoonshineStreamingEngine,
            StreamingModelParams,
@ -42,6 +43,7 @@ enum LoadedEngine {
    Moonshine(MoonshineEngine),
    MoonshineStreaming(MoonshineStreamingEngine),
    SenseVoice(SenseVoiceEngine),
+    GigaAM(GigaAMEngine),
 }

 #[derive(Clone)]
@ -165,6 +167,7 @@ impl TranscriptionManager {
                    LoadedEngine::Moonshine(ref mut e) => e.unload_model(),
                    LoadedEngine::MoonshineStreaming(ref mut e) => e.unload_model(),
                    LoadedEngine::SenseVoice(ref mut e) => e.unload_model(),
+                    LoadedEngine::GigaAM(ref mut e) => e.unload_model(),
                }
            }
            *engine = None; // Drop the engine to free memory
@ -346,6 +349,23 @@ impl TranscriptionManager {
                    })?;
                LoadedEngine::SenseVoice(engine)
            }
+            EngineType::GigaAM => {
+                let mut engine = GigaAMEngine::new();
+                engine.load_model(&model_path).map_err(|e| {
+                    let error_msg = format!("Failed to load gigaam model {}: {}", model_id, e);
+                    let _ = self.app_handle.emit(
+                        "model-state-changed",
+                        ModelStateEvent {
+                            event_type: "loading_failed".to_string(),
+                            model_id: Some(model_id.to_string()),
+                            model_name: Some(model_info.name.clone()),
+                            error: Some(error_msg.clone()),
+                        },
+                    );
+                    anyhow::anyhow!(error_msg)
+                })?;
+                LoadedEngine::GigaAM(engine)
+            }
        };

        // Update the current engine and model ID
@ -526,6 +546,9 @@ impl TranscriptionManager {
                                    anyhow::anyhow!("SenseVoice transcription failed: {}", e)
                                })
                        }
+                        LoadedEngine::GigaAM(gigaam_engine) => gigaam_engine
+                            .transcribe_samples(audio, None)
+                            .map_err(|e| anyhow::anyhow!("GigaAM transcription failed: {}", e)),
                    }
                },
            ));
--- a/src/bindings.ts
+++ b/src/bindings.ts
@ -744,7 +744,7 @@ export type AutoSubmitKey = "enter" | "ctrl_enter" | "cmd_enter"
 export type BindingResponse = { success: boolean; binding: ShortcutBinding | null; error: string | null }
 export type ClipboardHandling = "dont_modify" | "copy_to_clipboard"
 export type CustomSounds = { start: boolean; stop: boolean }
-export type EngineType = "Whisper" | "Parakeet" | "Moonshine" | "MoonshineStreaming" | "SenseVoice"
+export type EngineType = "Whisper" | "Parakeet" | "Moonshine" | "MoonshineStreaming" | "SenseVoice" | "GigaAM"
 export type HistoryEntry = { id: number; file_name: string; timestamp: number; saved: boolean; title: string; transcription_text: string; post_processed_text: string | null; post_process_prompt: string | null }
 /**
 * Result of changing keyboard implementation
--- a/src/i18n/locales/ar/translation.json
+++ b/src/i18n/locales/ar/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "سريع جداً. الصينية، الإنجليزية، اليابانية، الكورية، الكانتونية."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "التعرف على الكلام الروسي. سريع ودقيق."
      }
    },
    "errors": {
--- a/src/i18n/locales/cs/translation.json
+++ b/src/i18n/locales/cs/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Velmi rychlý. Čínština, angličtina, japonština, korejština, kantonština."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Rozpoznávání ruské řeči. Rychlé a přesné."
      }
    },
    "errors": {
--- a/src/i18n/locales/de/translation.json
+++ b/src/i18n/locales/de/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Sehr schnell. Chinesisch, Englisch, Japanisch, Koreanisch, Kantonesisch."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Russische Spracherkennung. Schnell und präzise."
      }
    },
    "errors": {
--- a/src/i18n/locales/en/translation.json
+++ b/src/i18n/locales/en/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Very fast. Chinese, English, Japanese, Korean, Cantonese."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Russian speech recognition. Fast and accurate."
      }
    },
    "errors": {
--- a/src/i18n/locales/es/translation.json
+++ b/src/i18n/locales/es/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Muy rápido. Chino, inglés, japonés, coreano, cantonés."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Reconocimiento de voz en ruso. Rápido y preciso."
      }
    },
    "errors": {
--- a/src/i18n/locales/fr/translation.json
+++ b/src/i18n/locales/fr/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Très rapide. Chinois, anglais, japonais, coréen, cantonais."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Reconnaissance vocale en russe. Rapide et précis."
      }
    },
    "errors": {
--- a/src/i18n/locales/it/translation.json
+++ b/src/i18n/locales/it/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Molto veloce. Cinese, inglese, giapponese, coreano, cantonese."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Riconoscimento vocale in russo. Veloce e preciso."
      }
    },
    "errors": {
--- a/src/i18n/locales/ja/translation.json
+++ b/src/i18n/locales/ja/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "非常に高速。中国語、英語、日本語、韓国語、広東語。"
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "ロシア語音声認識。高速かつ高精度。"
      }
    },
    "errors": {
--- a/src/i18n/locales/ko/translation.json
+++ b/src/i18n/locales/ko/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "매우 빠름. 중국어, 영어, 일본어, 한국어, 광둥어."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "러시아어 음성 인식. 빠르고 정확함."
      }
    },
    "errors": {
--- a/src/i18n/locales/pl/translation.json
+++ b/src/i18n/locales/pl/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Bardzo szybki. Chiński, angielski, japoński, koreański, kantoński."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Rozpoznawanie mowy rosyjskiej. Szybkie i dokładne."
      }
    },
    "errors": {
--- a/src/i18n/locales/pt/translation.json
+++ b/src/i18n/locales/pt/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Muito rápido. Chinês, inglês, japonês, coreano, cantonês."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Reconhecimento de voz em russo. Rápido e preciso."
      }
    },
    "errors": {
--- a/src/i18n/locales/ru/translation.json
+++ b/src/i18n/locales/ru/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Очень быстрый. Китайский, английский, японский, корейский, кантонский."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Распознавание русской речи. Быстро и точно."
      }
    },
    "errors": {
--- a/src/i18n/locales/tr/translation.json
+++ b/src/i18n/locales/tr/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Çok hızlı. Çince, İngilizce, Japonca, Korece, Kantonca."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Rusça konuşma tanıma. Hızlı ve doğru."
      }
    },
    "errors": {
--- a/src/i18n/locales/uk/translation.json
+++ b/src/i18n/locales/uk/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Дуже швидкий. Китайська, англійська, японська, корейська, кантонська."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Розпізнавання російського мовлення. Швидко і точно."
      }
    },
    "errors": {
--- a/src/i18n/locales/vi/translation.json
+++ b/src/i18n/locales/vi/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "Rất nhanh. Tiếng Trung, tiếng Anh, tiếng Nhật, tiếng Hàn, tiếng Quảng Đông."
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "Nhận dạng giọng nói tiếng Nga. Nhanh và chính xác."
      }
    },
    "errors": {
--- a/src/i18n/locales/zh-TW/translation.json
+++ b/src/i18n/locales/zh-TW/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "速度極快。支援中文、英語、日語、韓語、粵語。"
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "俄語語音辨識。快速且準確。"
      }
    },
    "errors": {
--- a/src/i18n/locales/zh/translation.json
+++ b/src/i18n/locales/zh/translation.json
@ -75,6 +75,10 @@
      "sense-voice-int8": {
        "name": "SenseVoice",
        "description": "非常快速。支持中文、英语、日语、韩语、粤语。"
+      },
+      "gigaam-v3-e2e-ctc": {
+        "name": "GigaAM v3",
+        "description": "俄语语音识别。快速且准确。"
      }
    },
    "errors": {