From 84268f3c4742c9e29dad7bd8693fa03d72a8e7a5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Geoffrey=20=E2=80=9CFrogeye=E2=80=9D=20Preud=27homme?=
 <geoffrey@frogeye.fr>
Date: Fri, 15 Nov 2024 22:41:26 +0100
Subject: [PATCH] Add whisperx

---
 hm/default.nix                      |  2 +-
 hm/{extra.nix => extra/default.nix} |  1 +
 hm/extra/whisperx.nix               | 37 +++++++++++++++++++++++++++++
 3 files changed, 39 insertions(+), 1 deletion(-)
 rename hm/{extra.nix => extra/default.nix} (95%)
 create mode 100644 hm/extra/whisperx.nix

diff --git a/hm/default.nix b/hm/default.nix
index 63e76a4..22917db 100644
--- a/hm/default.nix
+++ b/hm/default.nix
@@ -8,7 +8,7 @@
     ./common.nix
     ./desktop
     ./dev
-    ./extra.nix
+    ./extra
     ./gaming
     ./git
     ./gpg
diff --git a/hm/extra.nix b/hm/extra/default.nix
similarity index 95%
rename from hm/extra.nix
rename to hm/extra/default.nix
index 19e82b3..10580d7 100644
--- a/hm/extra.nix
+++ b/hm/extra/default.nix
@@ -21,6 +21,7 @@
 
       # Communication
       signal-desktop
+      (pkgs.callPackage ./whisperx.nix {}) # Transcribe voice messages
 
       # downloading
       # transmission TODO Collision if both transmissions are active?
diff --git a/hm/extra/whisperx.nix b/hm/extra/whisperx.nix
new file mode 100644
index 0000000..f9f3f69
--- /dev/null
+++ b/hm/extra/whisperx.nix
@@ -0,0 +1,37 @@
+{ pkgs ? import <nixpkgs> { } }:
+pkgs.python3Packages.buildPythonPackage {
+  pname = "whisperx";
+  version = "2024-08-19";
+  # pypi doesn't have the requirements.txt file, and it's required
+  src = pkgs.fetchFromGitHub {
+    owner = "m-bain";
+    repo = "whisperX";
+    rev = "9e3a9e0e38fcec1304e1784381059a0e2c670be5"; # git doesn't have tags
+    hash = "sha256-IVtn9fe/yi4+fbH57s9LoiREnMZ2nhEObp1a4R/7gHg=";
+  };
+  pyproject = true;
+  dependencies = [
+    pkgs.python3Packages.torch
+    pkgs.python3Packages.torchaudio
+    (pkgs.python3Packages.faster-whisper.overrideAttrs (old: {
+      # 1.0.2 is actually breaking APIs (requires hotwords argument)
+      src = pkgs.fetchFromGitHub {
+        owner = "SYSTRAN";
+        repo = "faster-whisper";
+        rev = "v1.0.0";
+        hash = "sha256-0fE8X1d6CgDrrHtRudksN/tIGRtBKMvoNwkSVyFNda4=";
+      };
+    }))
+    pkgs.python3Packages.transformers
+    pkgs.python3Packages.pyannote-audio # Not in the requirements.txt for some reason
+    pkgs.python3Packages.pandas
+    pkgs.python3Packages.nltk
+  ];
+  build-system = [
+    pkgs.python3Packages.setuptools
+  ];
+  pythonImportsCheck = [
+    "whisperx"
+  ];
+}
+