Feat/diarization v1 (#206)

* feat: initial support for diarization

* fix build errors

* bundle models

* more checks

* translations

* translations

* fix build

* fix windows build

* fix pre build with static crt

* misc

* fix pre build

* v2.4.0-beta.0

* fix pre build

* fix pre build

* misc

* fix pyannote rs

* faet: download models for diarization instead of embed

* fix speaker search

.gitignore

@@ -24,3 +24,4 @@ node_modules/
 .DS_Store
 venv/
 *.bin
+*.onnx

.vscode/settings.json

@@ -6,7 +6,7 @@
     },
     "files.eol": "\n",
     // Rust analyzer
-    "rust-analyzer.checkOnSave": false,
+    "rust-analyzer.checkOnSave": true,
     // "rust-analyzer.check.command": "clippy",
     "[rust]": {
         "editor.defaultFormatter": "rust-lang.rust-analyzer"

core/Cargo.toml

@@ -35,7 +35,7 @@ tokio = { version = "1.35.1", features = [
 ] }
 serde_json = { workspace = true }
 futures-util = "0.3.30"
-# sherpa-rs = { version = "0.1.4-beta.2", path = "../../sherpa-rs" }
+pyannote-rs = "0.2.2"
 
 [dev-dependencies]
 
@@ -47,13 +47,14 @@ glob = "0.3.1"
 
 [features]
 default = []
-diarize = ["sherpa-rs"]
-coreml = ["whisper-rs/coreml"]
+coreml = ["whisper-rs/coreml", "pyannote-rs/coreml"]
+directml = ["pyannote-rs/directml"]
 cuda = ["whisper-rs/cuda"]
 opencl = ["whisper-rs/opencl"]
 openblas = ["whisper-rs/openblas"]
 metal = ["whisper-rs/metal"]
 rocm = ["whisper-rs/hipblas"]
 
+
 [profile.test]
 inherits = "release"

core/build.rs

@@ -1,6 +1,6 @@
+use std::env;
 use std::path::{Path, PathBuf};
 use std::process::Command;
-use std::{env, fs};
 
 fn copy_file(src: &Path, dst: &Path) {
     if dst.exists() {

core/src/audio.rs

@@ -80,6 +80,10 @@ pub fn normalize(input: PathBuf, output: PathBuf) -> Result<()> {
     if !pid.wait()?.success() {
         bail!("unable to convert file")
     }
+
+    if !output.exists() {
+        bail!("seems like ffmpeg failed for some reason. output not exists")
+    }
     Ok(())
 }
 

core/src/config.rs

@@ -1,6 +1,5 @@
 use core::fmt;
 use serde::{Deserialize, Serialize};
-use std::path::PathBuf;
 use utoipa::ToSchema;
 
 #[derive(Deserialize, Serialize, ToSchema)]
@@ -18,12 +17,6 @@ pub struct TranscribeOptions {
     pub max_sentence_len: Option<i32>,
 }
 
-#[derive(Deserialize, Serialize, ToSchema)]
-pub struct DiarizeOptions {
-    pub vad_model_path: PathBuf,
-    pub speaker_id_model_path: PathBuf,
-}
-
 impl fmt::Debug for TranscribeOptions {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         let json_string = serde_json::to_string_pretty(self).map_err(|_| fmt::Error)?;