no render mac

k
fixed
2025-09-10 11:20:01 -06:00 · 2024-09-21 21:33:40 -04:00 · 2024-09-21 21:33:40 -04:00 · 2024-09-21 21:33:40 -04:00 · 2024-01-03 20:40:35 -07:00 · 2024-01-03 17:18:07 -07:00
14 changed files with 632 additions and 530 deletions
--- a/.gitmodules
+++ b/.gitmodules
@@ -1,3 +1,3 @@
-[submodule "rust-whisper.d/gitea-whisper-rs"]
+[submodule "gitea-whisper-rs"]
 	path = gitea-whisper-rs
 	url = https://gitea.inhome.blapointe.com/bel/whisper-rs.git
--- a/hotwords/transcript.sh
+++ b/hotwords/transcript.sh
@@ -13,6 +13,7 @@ if ! which rust-whisper-baked; then
 fi >&2

 cat <<EOF
+rust-whisper-baked --stream-device pulse_monitor --stream-step 16 --stream-retain 8 --stream-{head,tail}=0.25 2> /dev/null
 rust-whisper-baked --stream-device 'BlackHole 2ch' --stream-step 30 --stream-retain 1 --stream-{head,tail}=0.25 --threads 9 2> /dev/null
   | tee -a "$HOME/Sync/drawful/DnD/bdoob/__log.d/$(date +%Y.%m.%d).transcript.txt"
   | tee -a "$HOME/Sync/drawful/DnD/nessira.d/_log.d/$(date +%Y.%m.%d).transcript.txt"
--- a/listen-lib/src/lib.rs
+++ b/listen-lib/src/lib.rs
@@ -25,7 +25,11 @@ pub fn devices() -> Vec<String> {
 fn _devices() -> Result<Vec<cpal::Device>, String> {
   match cpal::default_host().devices() {
      Ok(devices) => Ok(devices.filter(|device| {
-         device.supported_input_configs().unwrap().count() > 0
+         let input_configs = device.supported_input_configs();
+         if !input_configs.is_ok() {
+            return false;
+         }
+         input_configs.unwrap().count() > 0
      }).collect()),
      Err(msg) => Err(format!("failed to get devices: {}", msg)),
   }
@@ -92,13 +96,22 @@ impl Listener {
         filter(|device| device.name().unwrap() == self.device_name).
         collect::<Vec<_>>();
      let device = devices.first().unwrap();
-      let cfg = device.supported_input_configs()
+      let mut sample_rate = 15_500;
+      let mut cfgs: Vec<_> = device.supported_input_configs()
         .unwrap()
         .filter(|x| x.sample_format() == cpal::SampleFormat::F32)
-         .filter(|x| x.min_sample_rate() >= cpal::SampleRate(15_500))
-         .nth(0)
+         .filter(|x| x.min_sample_rate() >= cpal::SampleRate(sample_rate))
+         .collect();
+      while cfgs.len() == 0 && sample_rate > 0 {
+         sample_rate /= 2;
+         cfgs = device.supported_input_configs()
            .unwrap()
-         .with_max_sample_rate();
+            .filter(|x| x.sample_format() == cpal::SampleFormat::F32)
+            .filter(|x| x.min_sample_rate() >= cpal::SampleRate(sample_rate))
+            .collect();
+      }
+      assert!(cfgs.len() > 0);
+      let cfg = cfgs[0].clone().with_max_sample_rate();

      let downsample_ratio = cfg.channels() as f32 * (cfg.sample_rate().0 as f32 / 16_000.0);
      let stream = device.build_input_stream(
--- a/rust-whisper-baked-lib/Cargo.lock
+++ b/rust-whisper-baked-lib/Cargo.lock
@@ -2,21 +2,6 @@
 # It is not intended for manual editing.
 version = 3

-[[package]]
-name = "addr2line"
-version = "0.21.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8a30b2e23b9e17a9f90641c7ab1549cd9b44f296d3ccbf309d2863cfe398a0cb"
-dependencies = [
- "gimli",
-]
-
-[[package]]
-name = "adler"
-version = "1.0.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f26201604c87b1e01bd3d98f8d5d9a8fcbb815e8cedb41ffccbeb4bf593a35fe"
-
 [[package]]
 name = "aho-corasick"
 version = "1.1.2"
@@ -26,21 +11,6 @@ dependencies = [
 "memchr",
 ]

-[[package]]
-name = "android-tzdata"
-version = "0.1.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e999941b234f3131b00bc13c22d06e8c5ff726d1b6318ac7eb276997bbb4fef0"
-
-[[package]]
-name = "android_system_properties"
-version = "0.1.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "819e7219dbd41043ac279b19830f2efc897156490d7fd6ea916720117ee66311"
-dependencies = [
- "libc",
-]
-
 [[package]]
 name = "anstream"
 version = "0.6.5"
@@ -89,27 +59,6 @@ dependencies = [
 "windows-sys",
 ]

-[[package]]
-name = "autocfg"
-version = "1.1.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d468802bab17cbc0cc575e9b053f41e72aa36bfa6b7f55e3529ffa43161b97fa"
-
-[[package]]
-name = "backtrace"
-version = "0.3.69"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2089b7e3f35b9dd2d0ed921ead4f6d318c27680d4a5bd167b3ee120edb105837"
-dependencies = [
- "addr2line",
- "cc",
- "cfg-if",
- "libc",
- "miniz_oxide",
- "object",
- "rustc-demangle",
-]
-
 [[package]]
 name = "bindgen"
 version = "0.68.1"
@@ -139,18 +88,6 @@ version = "2.4.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "327762f6e5a765692301e5bb513e0d9fef63be86bbc14528052b1cd3e6f03e07"

-[[package]]
-name = "bumpalo"
-version = "3.14.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7f30e7476521f6f8af1a1c4c0b8cc94f0bee37d91763d0ca2665f299b6cd8aec"
-
-[[package]]
-name = "byteorder"
-version = "1.5.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1fd0f2584146f6f2ef48085050886acf353beff7305ebd1ae69500e27c67f64b"
-
 [[package]]
 name = "cc"
 version = "1.0.83"
@@ -175,20 +112,6 @@ version = "1.0.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"

-[[package]]
-name = "chrono"
-version = "0.4.31"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7f2c685bad3eb3d45a01354cedb7d5faa66194d1d58ba6e267a8de788f79db38"
-dependencies = [
- "android-tzdata",
- "iana-time-zone",
- "js-sys",
- "num-traits",
- "wasm-bindgen",
- "windows-targets 0.48.5",
-]
-
 [[package]]
 name = "clang-sys"
 version = "1.6.1"
@@ -255,12 +178,6 @@ version = "1.0.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "acbf1af155f9b9ef647e42cdc158db4b64a1b61f743629225fde6f3e0be2a7c7"

-[[package]]
-name = "core-foundation-sys"
-version = "0.8.6"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "06ea2b9bc92be3c2baa9334a323ebca2d6f074ff852cd1d7b11064035cd3868f"
-
 [[package]]
 name = "either"
 version = "1.9.0"
@@ -283,12 +200,6 @@ version = "1.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "42703706b716c37f96a77aea830392ad231f44c9e9a67872fa5548707e11b11c"

-[[package]]
-name = "gimli"
-version = "0.28.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4271d37baee1b8c7e4b708028c57d816cf9d2434acb33a549475f78c181f6253"
-
 [[package]]
 name = "glob"
 version = "0.3.1"
@@ -310,38 +221,6 @@ dependencies = [
 "windows-sys",
 ]

-[[package]]
-name = "iana-time-zone"
-version = "0.1.58"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8326b86b6cff230b97d0d312a6c40a60726df3332e721f72a1b035f451663b20"
-dependencies = [
- "android_system_properties",
- "core-foundation-sys",
- "iana-time-zone-haiku",
- "js-sys",
- "wasm-bindgen",
- "windows-core",
-]
-
-[[package]]
-name = "iana-time-zone-haiku"
-version = "0.1.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f31827a206f56af32e590ba56d5d2d085f558508192593743f16b2306495269f"
-dependencies = [
- "cc",
-]
-
-[[package]]
-name = "js-sys"
-version = "0.3.66"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cee9c64da59eae3b50095c18d3e74f8b73c0b86d2792824ff01bbce68ba229ca"
-dependencies = [
- "wasm-bindgen",
-]
-
 [[package]]
 name = "lazy_static"
 version = "1.4.0"
@@ -394,15 +273,6 @@ version = "0.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "68354c5c6bd36d73ff3feceb05efa59b6acb7626617f4962be322a825e61f79a"

-[[package]]
-name = "miniz_oxide"
-version = "0.7.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e7810e0be55b428ada41041c41f32c9f1a42817901b4ccf45fa3d4b6561e74c7"
-dependencies = [
- "adler",
-]
-
 [[package]]
 name = "nom"
 version = "7.1.3"
@@ -413,24 +283,6 @@ dependencies = [
 "minimal-lexical",
 ]

-[[package]]
-name = "num-traits"
-version = "0.2.17"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "39e3200413f237f41ab11ad6d161bc7239c84dcb631773ccd7de3dfe4b5c267c"
-dependencies = [
- "autocfg",
-]
-
-[[package]]
-name = "object"
-version = "0.32.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9cf5f9dd3933bd50a9e1f149ec995f39ae2c496d31fd772c1fd45ebc27e902b0"
-dependencies = [
- "memchr",
-]
-
 [[package]]
 name = "once_cell"
 version = "1.19.0"
@@ -443,12 +295,6 @@ version = "0.1.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "19b17cddbe7ec3f8bc800887bab5e717348c95ea2ca0b1bf0837fb964dc67099"

-[[package]]
-name = "pin-project-lite"
-version = "0.2.13"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8afb450f006bf6385ca15ef45d71d2288452bc3683ce2e2cacc0d18e4be60b58"
-
 [[package]]
 name = "prettyplease"
 version = "0.2.15"
@@ -523,20 +369,11 @@ dependencies = [
 name = "rust-whisper-lib"
 version = "0.1.0"
 dependencies = [
- "byteorder",
- "chrono",
 "clap",
- "tokio",
 "wav",
 "whisper-rs",
 ]

-[[package]]
-name = "rustc-demangle"
-version = "0.1.23"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d626bb9dae77e28219937af045c257c28bfd3f69333c512553507f5f9798cb76"
-
 [[package]]
 name = "rustc-hash"
 version = "1.1.0"
@@ -579,16 +416,6 @@ dependencies = [
 "unicode-ident",
 ]

-[[package]]
-name = "tokio"
-version = "1.35.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c89b4efa943be685f629b149f53829423f8f5531ea21249408e8e2f8671ec104"
-dependencies = [
- "backtrace",
- "pin-project-lite",
-]
-
 [[package]]
 name = "unicode-ident"
 version = "1.0.12"
@@ -601,60 +428,6 @@ version = "0.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "711b9620af191e0cdc7468a8d14e709c3dcdb115b36f838e601583af800a370a"

-[[package]]
-name = "wasm-bindgen"
-version = "0.2.89"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0ed0d4f68a3015cc185aff4db9506a015f4b96f95303897bfa23f846db54064e"
-dependencies = [
- "cfg-if",
- "wasm-bindgen-macro",
-]
-
-[[package]]
-name = "wasm-bindgen-backend"
-version = "0.2.89"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1b56f625e64f3a1084ded111c4d5f477df9f8c92df113852fa5a374dbda78826"
-dependencies = [
- "bumpalo",
- "log",
- "once_cell",
- "proc-macro2",
- "quote",
- "syn",
- "wasm-bindgen-shared",
-]
-
-[[package]]
-name = "wasm-bindgen-macro"
-version = "0.2.89"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0162dbf37223cd2afce98f3d0785506dcb8d266223983e4b5b525859e6e182b2"
-dependencies = [
- "quote",
- "wasm-bindgen-macro-support",
-]
-
-[[package]]
-name = "wasm-bindgen-macro-support"
-version = "0.2.89"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f0eb82fcb7930ae6219a7ecfd55b217f5f0893484b7a13022ebb2b2bf20b5283"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn",
- "wasm-bindgen-backend",
- "wasm-bindgen-shared",
-]
-
-[[package]]
-name = "wasm-bindgen-shared"
-version = "0.2.89"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7ab9b36309365056cd639da3134bf87fa8f3d86008abf99e612384a6eecd459f"
-
 [[package]]
 name = "wav"
 version = "1.0.0"
@@ -715,37 +488,13 @@ version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"

-[[package]]
-name = "windows-core"
-version = "0.51.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f1f8cf84f35d2db49a46868f947758c7a1138116f7fac3bc844f43ade1292e64"
-dependencies = [
- "windows-targets 0.48.5",
-]
-
 [[package]]
 name = "windows-sys"
 version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "282be5f36a8ce781fad8c8ae18fa3f9beff57ec1b52cb3de0789201425d9a33d"
 dependencies = [
- "windows-targets 0.52.0",
-]
-
-[[package]]
-name = "windows-targets"
-version = "0.48.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9a2fa6e2155d7247be68c096456083145c183cbbbc2764150dda45a87197940c"
-dependencies = [
- "windows_aarch64_gnullvm 0.48.5",
- "windows_aarch64_msvc 0.48.5",
- "windows_i686_gnu 0.48.5",
- "windows_i686_msvc 0.48.5",
- "windows_x86_64_gnu 0.48.5",
- "windows_x86_64_gnullvm 0.48.5",
- "windows_x86_64_msvc 0.48.5",
+ "windows-targets",
 ]

 [[package]]
@@ -754,93 +503,51 @@ version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "8a18201040b24831fbb9e4eb208f8892e1f50a37feb53cc7ff887feb8f50e7cd"
 dependencies = [
- "windows_aarch64_gnullvm 0.52.0",
- "windows_aarch64_msvc 0.52.0",
- "windows_i686_gnu 0.52.0",
- "windows_i686_msvc 0.52.0",
- "windows_x86_64_gnu 0.52.0",
- "windows_x86_64_gnullvm 0.52.0",
- "windows_x86_64_msvc 0.52.0",
+ "windows_aarch64_gnullvm",
+ "windows_aarch64_msvc",
+ "windows_i686_gnu",
+ "windows_i686_msvc",
+ "windows_x86_64_gnu",
+ "windows_x86_64_gnullvm",
+ "windows_x86_64_msvc",
 ]

-[[package]]
-name = "windows_aarch64_gnullvm"
-version = "0.48.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2b38e32f0abccf9987a4e3079dfb67dcd799fb61361e53e2882c3cbaf0d905d8"
-
 [[package]]
 name = "windows_aarch64_gnullvm"
 version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "cb7764e35d4db8a7921e09562a0304bf2f93e0a51bfccee0bd0bb0b666b015ea"

-[[package]]
-name = "windows_aarch64_msvc"
-version = "0.48.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dc35310971f3b2dbbf3f0690a219f40e2d9afcf64f9ab7cc1be722937c26b4bc"
-
 [[package]]
 name = "windows_aarch64_msvc"
 version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "bbaa0368d4f1d2aaefc55b6fcfee13f41544ddf36801e793edbbfd7d7df075ef"

-[[package]]
-name = "windows_i686_gnu"
-version = "0.48.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a75915e7def60c94dcef72200b9a8e58e5091744960da64ec734a6c6e9b3743e"
-
 [[package]]
 name = "windows_i686_gnu"
 version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "a28637cb1fa3560a16915793afb20081aba2c92ee8af57b4d5f28e4b3e7df313"

-[[package]]
-name = "windows_i686_msvc"
-version = "0.48.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8f55c233f70c4b27f66c523580f78f1004e8b5a8b659e05a4eb49d4166cca406"
-
 [[package]]
 name = "windows_i686_msvc"
 version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "ffe5e8e31046ce6230cc7215707b816e339ff4d4d67c65dffa206fd0f7aa7b9a"

-[[package]]
-name = "windows_x86_64_gnu"
-version = "0.48.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "53d40abd2583d23e4718fddf1ebec84dbff8381c07cae67ff7768bbf19c6718e"
-
 [[package]]
 name = "windows_x86_64_gnu"
 version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3d6fa32db2bc4a2f5abeacf2b69f7992cd09dca97498da74a151a3132c26befd"

-[[package]]
-name = "windows_x86_64_gnullvm"
-version = "0.48.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0b7b52767868a23d5bab768e390dc5f5c55825b6d30b86c844ff2dc7414044cc"
-
 [[package]]
 name = "windows_x86_64_gnullvm"
 version = "0.52.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1a657e1e9d3f514745a572a6846d3c7aa7dbe1658c056ed9c3344c4109a6949e"

-[[package]]
-name = "windows_x86_64_msvc"
-version = "0.48.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ed94fce61571a4006852b7389a063ab983c02eb1bb37b47f8272ce92d06d9538"
-
 [[package]]
 name = "windows_x86_64_msvc"
 version = "0.52.0"
--- a/rust-whisper-baked-lib/src/lib.rs
+++ b/rust-whisper-baked-lib/src/lib.rs
@@ -6,7 +6,7 @@ pub fn channel<F>(
   stream: std::sync::mpsc::Receiver<Vec<f32>>,
 ) where F: FnMut(Result<rust_whisper_lib::Transcribed, String>) + Send + 'static {
   flags.model_path = None;
-   flags.model_buffer = Some(include_bytes!("../../models/ggml-tiny.en.bin").to_vec());
+   flags.model_buffer = Some(get_fast());
   rust_whisper_lib::channel(flags.clone(), handler_fn, stream);
 }

@@ -15,7 +15,27 @@ pub fn wav<F>(
   handler_fn: F
 ) where F: FnMut(Result<rust_whisper_lib::Transcribed, String>) + Send + 'static {
   flags.model_path = None;
-   flags.model_buffer = Some(include_bytes!("../../models/ggml-distil-medium.en.bin").to_vec());
+   flags.model_buffer = Some(get_good());
   rust_whisper_lib::wav(flags.clone(), handler_fn, flags.wav.unwrap());
 }

+pub fn wav_channel<F>(
+   mut flags: rust_whisper_lib::Flags,
+   handler_fn: F
+) where F: FnMut(Result<rust_whisper_lib::Transcribed, String>) + Send + 'static {
+   flags.model_path = None;
+   flags.model_buffer = Some(get_good());
+   rust_whisper_lib::wav_channel(flags, handler_fn);
+}
+
+pub fn f32_from_wav_file(path: &String) -> Result<Vec<f32>, String> {
+   rust_whisper_lib::f32_from_wav_file(path)
+}
+
+fn get_fast() -> Vec<u8> {
+   include_bytes!("../../models/ggml-small.en.bin").to_vec()
+}
+
+fn get_good() -> Vec<u8> {
+   include_bytes!("../../models/ggml-distil-medium.en.bin").to_vec()
+}
--- a/rust-whisper-baked/Cargo.lock
+++ b/rust-whisper-baked/Cargo.lock
@@ -2,21 +2,6 @@
 # It is not intended for manual editing.
 version = 3

-[[package]]
-name = "addr2line"
-version = "0.21.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8a30b2e23b9e17a9f90641c7ab1549cd9b44f296d3ccbf309d2863cfe398a0cb"
-dependencies = [
- "gimli",
-]
-
-[[package]]
-name = "adler"
-version = "1.0.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f26201604c87b1e01bd3d98f8d5d9a8fcbb815e8cedb41ffccbeb4bf593a35fe"
-
 [[package]]
 name = "aho-corasick"
 version = "1.1.2"
@@ -48,21 +33,6 @@ dependencies = [
 "pkg-config",
 ]

-[[package]]
-name = "android-tzdata"
-version = "0.1.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e999941b234f3131b00bc13c22d06e8c5ff726d1b6318ac7eb276997bbb4fef0"
-
-[[package]]
-name = "android_system_properties"
-version = "0.1.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "819e7219dbd41043ac279b19830f2efc897156490d7fd6ea916720117ee66311"
-dependencies = [
- "libc",
-]
-
 [[package]]
 name = "anstream"
 version = "0.6.5"
@@ -117,21 +87,6 @@ version = "1.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d468802bab17cbc0cc575e9b053f41e72aa36bfa6b7f55e3529ffa43161b97fa"

-[[package]]
-name = "backtrace"
-version = "0.3.69"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2089b7e3f35b9dd2d0ed921ead4f6d318c27680d4a5bd167b3ee120edb105837"
-dependencies = [
- "addr2line",
- "cc",
- "cfg-if",
- "libc",
- "miniz_oxide",
- "object",
- "rustc-demangle",
-]
-
 [[package]]
 name = "bindgen"
 version = "0.68.1"
@@ -193,12 +148,6 @@ version = "3.14.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7f30e7476521f6f8af1a1c4c0b8cc94f0bee37d91763d0ca2665f299b6cd8aec"

-[[package]]
-name = "byteorder"
-version = "1.5.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1fd0f2584146f6f2ef48085050886acf353beff7305ebd1ae69500e27c67f64b"
-
 [[package]]
 name = "bytes"
 version = "1.5.0"
@@ -236,20 +185,6 @@ version = "1.0.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"

-[[package]]
-name = "chrono"
-version = "0.4.31"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7f2c685bad3eb3d45a01354cedb7d5faa66194d1d58ba6e267a8de788f79db38"
-dependencies = [
- "android-tzdata",
- "iana-time-zone",
- "js-sys",
- "num-traits",
- "wasm-bindgen",
- "windows-targets 0.48.5",
-]
-
 [[package]]
 name = "clang-sys"
 version = "1.6.1"
@@ -411,12 +346,6 @@ version = "1.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "42703706b716c37f96a77aea830392ad231f44c9e9a67872fa5548707e11b11c"

-[[package]]
-name = "gimli"
-version = "0.28.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4271d37baee1b8c7e4b708028c57d816cf9d2434acb33a549475f78c181f6253"
-
 [[package]]
 name = "glob"
 version = "0.3.1"
@@ -444,29 +373,6 @@ dependencies = [
 "windows-sys",
 ]

-[[package]]
-name = "iana-time-zone"
-version = "0.1.58"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8326b86b6cff230b97d0d312a6c40a60726df3332e721f72a1b035f451663b20"
-dependencies = [
- "android_system_properties",
- "core-foundation-sys",
- "iana-time-zone-haiku",
- "js-sys",
- "wasm-bindgen",
- "windows-core",
-]
-
-[[package]]
-name = "iana-time-zone-haiku"
-version = "0.1.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f31827a206f56af32e590ba56d5d2d085f558508192593743f16b2306495269f"
-dependencies = [
- "cc",
-]
-
 [[package]]
 name = "indexmap"
 version = "2.1.0"
@@ -477,6 +383,12 @@ dependencies = [
 "hashbrown",
 ]

+[[package]]
+name = "itoa"
+version = "1.0.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b1a46d1a171d865aa5f83f92695765caa047a9b4cbae2cbf37dbd613a793fd4c"
+
 [[package]]
 name = "jni"
 version = "0.19.0"
@@ -608,15 +520,6 @@ version = "0.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "68354c5c6bd36d73ff3feceb05efa59b6acb7626617f4962be322a825e61f79a"

-[[package]]
-name = "miniz_oxide"
-version = "0.7.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e7810e0be55b428ada41041c41f32c9f1a42817901b4ccf45fa3d4b6561e74c7"
-dependencies = [
- "adler",
-]
-
 [[package]]
 name = "ndk"
 version = "0.7.0"
@@ -708,15 +611,6 @@ dependencies = [
 "syn 1.0.109",
 ]

-[[package]]
-name = "object"
-version = "0.32.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9cf5f9dd3933bd50a9e1f149ec995f39ae2c496d31fd772c1fd45ebc27e902b0"
-dependencies = [
- "memchr",
-]
-
 [[package]]
 name = "oboe"
 version = "0.5.0"
@@ -775,12 +669,6 @@ version = "0.1.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "19b17cddbe7ec3f8bc800887bab5e717348c95ea2ca0b1bf0837fb964dc67099"

-[[package]]
-name = "pin-project-lite"
-version = "0.2.13"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8afb450f006bf6385ca15ef45d71d2288452bc3683ce2e2cacc0d18e4be60b58"
-
 [[package]]
 name = "pkg-config"
 version = "0.3.27"
@@ -875,14 +763,26 @@ version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b9b1a3d5f46d53f4a3478e2be4a5a5ce5108ea58b100dcd139830eae7f79a3a1"

+[[package]]
+name = "rust-stemmers"
+version = "1.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e46a2036019fdb888131db7a4c847a1063a7493f971ed94ea82c67eada63ca54"
+dependencies = [
+ "serde",
+ "serde_derive",
+]
+
 [[package]]
 name = "rust-whisper-baked"
 version = "0.1.0"
 dependencies = [
 "clap",
 "listen-lib",
+ "rust-stemmers",
 "rust-whisper-baked-lib",
 "rust-whisper-lib",
+ "stop-words",
 ]

 [[package]]
@@ -896,20 +796,11 @@ dependencies = [
 name = "rust-whisper-lib"
 version = "0.1.0"
 dependencies = [
- "byteorder",
- "chrono",
 "clap",
- "tokio",
 "wav",
 "whisper-rs",
 ]

-[[package]]
-name = "rustc-demangle"
-version = "0.1.23"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d626bb9dae77e28219937af045c257c28bfd3f69333c512553507f5f9798cb76"
-
 [[package]]
 name = "rustc-hash"
 version = "1.1.0"
@@ -929,6 +820,12 @@ dependencies = [
 "windows-sys",
 ]

+[[package]]
+name = "ryu"
+version = "1.0.16"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f98d2aa92eebf49b69786be48e4477826b256916e84a57ff2a4f21923b48eb4c"
+
 [[package]]
 name = "same-file"
 version = "1.0.6"
@@ -944,6 +841,37 @@ version = "1.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "94143f37725109f92c262ed2cf5e59bce7498c01bcc1502d7b9afe439a4e9f49"

+[[package]]
+name = "serde"
+version = "1.0.193"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "25dd9975e68d0cb5aa1120c288333fc98731bd1dd12f561e468ea4728c042b89"
+dependencies = [
+ "serde_derive",
+]
+
+[[package]]
+name = "serde_derive"
+version = "1.0.193"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "43576ca501357b9b071ac53cdc7da8ef0cbd9493d8df094cd821777ea6e894d3"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.41",
+]
+
+[[package]]
+name = "serde_json"
+version = "1.0.109"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cb0652c533506ad7a2e353cce269330d6afd8bdfb6d75e0ace5b35aacbd7b9e9"
+dependencies = [
+ "itoa",
+ "ryu",
+ "serde",
+]
+
 [[package]]
 name = "shlex"
 version = "1.2.0"
@@ -975,6 +903,15 @@ version = "1.11.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "4dccd0940a2dcdf68d092b8cbab7dc0ad8fa938bf95787e1b916b0e3d0e8e970"

+[[package]]
+name = "stop-words"
+version = "0.8.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8500024d809de02ecbf998472b7bed3c4fca380df2be68917f6a473bdb28ddcc"
+dependencies = [
+ "serde_json",
+]
+
 [[package]]
 name = "strsim"
 version = "0.10.0"
@@ -1023,16 +960,6 @@ dependencies = [
 "syn 2.0.41",
 ]

-[[package]]
-name = "tokio"
-version = "1.35.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c89b4efa943be685f629b149f53829423f8f5531ea21249408e8e2f8671ec104"
-dependencies = [
- "backtrace",
- "pin-project-lite",
-]
-
 [[package]]
 name = "toml_datetime"
 version = "0.6.5"
@@ -1226,15 +1153,6 @@ dependencies = [
 "windows-targets 0.42.2",
 ]

-[[package]]
-name = "windows-core"
-version = "0.51.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f1f8cf84f35d2db49a46868f947758c7a1138116f7fac3bc844f43ade1292e64"
-dependencies = [
- "windows-targets 0.48.5",
-]
-
 [[package]]
 name = "windows-sys"
 version = "0.52.0"
--- a/rust-whisper-baked/Cargo.toml
+++ b/rust-whisper-baked/Cargo.toml
@@ -10,3 +10,5 @@ rust-whisper-lib = { path = "../rust-whisper-lib" }
 rust-whisper-baked-lib = { path = "../rust-whisper-baked-lib" }
 listen-lib = { path = "../listen-lib" }
 clap = { version = "4.4.10", features = ["derive"] }
+stop-words = "0.8.0"
+rust-stemmers = "1.2.0"
--- a/rust-whisper-baked/src/main.rs
+++ b/rust-whisper-baked/src/main.rs
@@ -1,21 +1,70 @@
-use rust_whisper_lib;
-use rust_whisper_baked_lib;
 use clap::Parser;
 use listen_lib;
+use rust_whisper_baked_lib;
+use rust_whisper_lib;
 use std::thread;

 fn main() {
    let flags = rust_whisper_lib::Flags::parse();
+    match flags.wav.clone() {
+        Some(_) => wav_channel(flags),
+        None => channel(flags),
+    };
+}
+
+fn wav_channel(flags: rust_whisper_lib::Flags) {
+    let mut w = new_destutterer();
+    rust_whisper_baked_lib::wav_channel(
+        flags.clone(),
+        move |result: Result<rust_whisper_lib::Transcribed, String>| {
+            match result {
+                Ok(transcribed) => {
+                    let s = w.step(transcribed.to_string());
+                    println!("{}", s);
+                }
+                Err(msg) => {
+                    eprintln!("error: {}", msg);
+                }
+            };
+        },
+    );
+}
+
+fn wav(flags: rust_whisper_lib::Flags, _path: String) {
+    let mut w = new_destutterer();
+    rust_whisper_baked_lib::wav(
+        flags,
+        move |result: Result<rust_whisper_lib::Transcribed, String>| {
+            match result {
+                Ok(transcribed) => {
+                    let s = w.step(transcribed.to_string());
+                    println!("{}", s);
+                }
+                Err(msg) => {
+                    eprintln!("error: {}", msg);
+                }
+            };
+        },
+    );
+}
+
+fn channel(flags: rust_whisper_lib::Flags) {
    let (send, recv) = std::sync::mpsc::sync_channel(100);

    eprintln!("rust whisper baked lib channel...");
    thread::spawn(move || {
+        let mut w = new_destutterer();
        rust_whisper_baked_lib::channel(
            flags.clone(),
-         |result: Result<rust_whisper_lib::Transcribed, String>| {
+            move |result: Result<rust_whisper_lib::Transcribed, String>| {
                match result {
-               Ok(transcribed) => { println!("{}", transcribed.to_string()); },
-               Err(msg) => { eprintln!("error: {}", msg); },
+                    Ok(transcribed) => {
+                        let s = w.step(transcribed.to_string());
+                        println!("{}", s);
+                    }
+                    Err(msg) => {
+                        eprintln!("error: {}", msg);
+                    }
                };
            },
            recv,
@@ -26,17 +75,25 @@ fn main() {
    let flags = rust_whisper_lib::Flags::parse();
    match flags.stream_device {
        Some(device_name) => {
-         if device_name == "" {
+            eprintln!("with device ({}) '{}'", device_name.len(), &device_name);
+            if device_name.len() == 0 {
+                let mut i = 0;
                for device in listen_lib::devices() {
-               eprintln!("{}", device);
+                    eprintln!("[{}] {}", i, device);
+                    i += 1;
                }
+                eprintln!("found {} devices", i);
            } else {
-            listen_lib::main_with(|data| {
+                listen_lib::main_with(
+                    |data| {
                        send.send(data).unwrap();
-            }, device_name);
-         }
                    },
+                    device_name,
+                );
+            }
+        }
        None => {
+            eprintln!("without any device");
            listen_lib::main(|data| {
                send.send(data).unwrap();
            });
@@ -44,3 +101,184 @@ fn main() {
    }
    eprintln!("/listen lib main...");
 }
+
+struct Destutterer {
+    prev: Words,
+}
+
+fn new_destutterer() -> Destutterer {
+    Destutterer { prev: new_words() }
+}
+
+impl Destutterer {
+    fn step(&mut self, next: String) -> String {
+        if next.len() == 0 {
+            return next;
+        }
+
+        let next_words = Words::from_string(next.clone());
+        let mut n = self
+            .prev
+            .comparable_len()
+            .clamp(0, next_words.comparable_len());
+        //println!("n={} prev='{:?}' next='{:?}'", n, self.prev.to_comparable_words(), next_words.to_comparable_words());
+        while n > 0 {
+            let (prev_s, _) = self.prev.last_n_comparable_to_string(n);
+            let (next_s, next_idx) = next_words.first_n_comparable_to_string(n);
+            if prev_s == next_s {
+                self.prev = next_words;
+                return self.prev.skip(next_idx + 1).to_string();
+            }
+            n -= 1;
+        }
+        self.prev = next_words;
+        self.prev.to_string()
+    }
+}
+
+#[derive(Clone, Debug)]
+struct Words {
+    raw: Vec<String>,
+}
+
+fn new_words() -> Words {
+    Words { raw: vec![] }
+}
+
+impl Words {
+    fn from_string(s: String) -> Words {
+        let mut result = Words { raw: vec![] };
+        for word in s.split(" ") {
+            let word = word.trim();
+            if word.len() > 0 {
+                result.raw.push(word.to_string());
+            }
+        }
+        result
+    }
+
+    fn skip(&self, n: usize) -> Words {
+        Words {
+            raw: self.raw.iter().skip(n).map(|x| x.clone()).collect(),
+        }
+    }
+
+    fn last_n_comparable_to_string(&self, n: usize) -> (String, usize) {
+        let v = self.to_comparable_words();
+        let v = v[(v.len() - n).clamp(0, v.len())..].to_vec();
+        return (
+            v.iter()
+                .map(|x| x.s.clone().unwrap())
+                .collect::<Vec<String>>()
+                .join(" "),
+            v[0].idx,
+        );
+    }
+
+    fn first_n_comparable_to_string(&self, n: usize) -> (String, usize) {
+        let v = self.to_comparable_words();
+        let v = v[0..n.clamp(0, v.len())].to_vec();
+        return (
+            v.iter()
+                .map(|x| x.s.clone().unwrap())
+                .collect::<Vec<String>>()
+                .join(" "),
+            v[v.len() - 1].idx,
+        );
+    }
+
+    fn comparable_len(&self) -> usize {
+        self.to_comparable_words().len()
+    }
+
+    fn to_comparable_words(&self) -> Vec<Word> {
+        self.to_words()
+            .iter()
+            .filter(|x| x.s.is_some())
+            .map(|x| x.clone())
+            .collect()
+    }
+
+    fn to_words(&self) -> Vec<Word> {
+        let skips = stop_words::get("en");
+        let stemmer = rust_stemmers::Stemmer::create(rust_stemmers::Algorithm::English);
+        let strs = self
+            .raw
+            .iter()
+            .map(|w| w.to_lowercase())
+            .map(|w| {
+                w.chars()
+                    .filter(|c| c.is_ascii_alphanumeric())
+                    .collect::<String>()
+            })
+            .map(|w| stemmer.stem(&w).into_owned())
+            .collect::<Vec<String>>();
+        let mut result = vec![];
+        for i in 0..strs.len() {
+            result.push(Word {
+                s: match skips.contains(&strs[i]) {
+                    true => None,
+                    false => Some(strs[i].clone()),
+                },
+                idx: i as usize,
+            });
+        }
+        result
+    }
+
+    fn to_string(&self) -> String {
+        self.raw
+            .iter()
+            .map(|x| x.clone())
+            .collect::<Vec<String>>()
+            .join(" ")
+    }
+}
+
+#[derive(Debug, Clone)]
+struct Word {
+    s: Option<String>,
+    idx: usize,
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_destutterer_stop_words() {
+        let mut w = new_destutterer();
+        assert_eq!(
+            "welcome to the internet".to_string(),
+            w.step("welcome to the internet".to_string())
+        );
+        assert_eq!(
+            "have a look around".to_string(),
+            w.step("welcome to the a internet; have a look around".to_string())
+        );
+    }
+
+    #[test]
+    fn test_destutterer_punctuation() {
+        let mut w = new_destutterer();
+        assert_eq!(
+            "cat, dog. cow? moose!".to_string(),
+            w.step("cat, dog. cow? moose!".to_string())
+        );
+        assert_eq!(
+            "elephant! fez gator".to_string(),
+            w.step("moose, elephant! fez gator".to_string())
+        );
+        assert_eq!("hij".to_string(), w.step("fez gator hij".to_string()));
+    }
+
+    #[test]
+    fn test_destutterer_basic() {
+        let mut w = new_destutterer();
+        assert_eq!(
+            "cat dog cow".to_string(),
+            w.step(" cat dog cow ".to_string())
+        );
+        assert_eq!("moose".to_string(), w.step(" dog cow moose  ".to_string()));
+    }
+}
--- a/rust-whisper-lib/src/lib.rs
+++ b/rust-whisper-lib/src/lib.rs
@@ -14,13 +14,13 @@ pub struct Flags {
   #[arg(long, default_value = "8")]
   pub threads: i32,

-   #[arg(long, default_value = "5")]
+   #[arg(long, default_value = "8")]
   pub stream_step: u64,
-   #[arg(long, default_value = "0.6")]
+   #[arg(long, default_value = "4.0")]
   pub stream_retain: f32,
-   #[arg(long, default_value = "0.3")]
+   #[arg(long, default_value = "2.0")]
   pub stream_head: f32,
-   #[arg(long, default_value = "0.3")]
+   #[arg(long, default_value = "0.0")]
   pub stream_tail: f32,

   #[arg(long, default_value = "false")]
@@ -44,7 +44,7 @@ pub fn wav<F>(flags: Flags, handler_fn: F, wav_path: String) where F: FnMut(Resu
   w.transcribe(&f32_from_wav_file(&wav_path).unwrap())
 }

-fn f32_from_wav_file(path: &String) -> Result<Vec<f32>, String> {
+pub fn f32_from_wav_file(path: &String) -> Result<Vec<f32>, String> {
   let f = std::fs::File::open(path);
   if let Some(err) = f.as_ref().err() {
      return Err(format!("failed to open wav file: {}", err));
@@ -72,7 +72,30 @@ fn f32_from_wav_file(path: &String) -> Result<Vec<f32>, String> {
   }
 }

-pub fn channel<F>(flags: Flags, handler_fn: F, stream: std::sync::mpsc::Receiver<Vec<f32>>) where F: FnMut(Result<Transcribed, String>) + Send + 'static {
+pub fn wav_channel<F>(flags: Flags, handler_fn: F) where F: FnMut(Result<Transcribed, String>) + Send + 'static {
+   let path = flags.wav.as_ref().unwrap();
+   let mut audio = f32_from_wav_file(&path).unwrap();
+   let mut iter = vec![];
+   let n = audio.len() / match audio.len() % 100 {
+      0 => 100,
+      _ => 99,
+   };
+   for _ in 0..100 {
+      iter.push(audio.drain(0..n.clamp(0, audio.len())).collect());
+   }
+   let (fin_send, fin_recv) = std::sync::mpsc::sync_channel::<Option<i32>>(1);
+   channel_and_close(flags.clone(), handler_fn, iter, move || { fin_send.send(None).unwrap(); });
+   match fin_recv.recv() {
+      Ok(_) => {},
+      Err(x) => panic!("failed to receive: {}", x),
+   };
+}
+
+pub fn channel<F, I>(flags: Flags, handler_fn: F, stream: I) where F: FnMut(Result<Transcribed, String>) + Send + 'static, I: IntoIterator<Item = Vec<f32>> {
+   channel_and_close(flags, handler_fn, stream, || {});
+}
+
+fn channel_and_close<F, I, G>(flags: Flags, handler_fn: F, stream: I, mut close_fn: G) where F: FnMut(Result<Transcribed, String>) + Send + 'static, I: IntoIterator<Item = Vec<f32>>, G: FnMut() + Send + 'static {
   let w = new_service(
      flags.model_path,
      flags.model_buffer,
@@ -87,7 +110,7 @@ pub fn channel<F>(flags: Flags, handler_fn: F, stream: std::sync::mpsc::Receiver
      false => {},
   };
   let mut buffer = vec![];
-   for data in stream.iter() {
+   for data in stream {
      data.iter().for_each(|x| buffer.push(*x));
      if buffer.len() >= (flags.stream_step * 16_000) as usize {
         w.transcribe_async(&buffer).unwrap();
@@ -112,6 +135,10 @@ pub fn channel<F>(flags: Flags, handler_fn: F, stream: std::sync::mpsc::Receiver
         buffer.truncate(stream_retain);
      }
   }
+   if buffer.len() > 0 {
+      w.transcribe(&buffer);
+   }
+   close_fn();
 }

 struct Service {
@@ -203,8 +230,10 @@ impl Impl {
      let result = whispered
         .after(&(self.stream_head * 100.0))
         .before(&(self.stream_tail * 100.0));
+      if result.to_string().trim().len() > 0 {
         (self.handler_fn.as_mut().unwrap())(Ok(result));
      }
+   }

   fn on_error(&mut self, msg: String) {
      (self.handler_fn.as_mut().unwrap())(Err(format!("failed to transcribe: {}", &msg)));
@@ -263,19 +292,6 @@ impl Engine {
   }
 }

-struct Engine2 {
-}
-
-fn new_engine_2() -> Result<Engine2, String> {
-   Ok(Engine2{})
-}
-
-impl Engine2 {
-   fn transcribe(&self, data: &Vec<f32>) -> Result<Transcribed, String> {
-      Err(format!("not impl"))
-   }
-}
-
 struct ATranscribe {
   data: Vec<f32>,
   ack: Option<std::sync::mpsc::SyncSender<bool>>,
@@ -352,6 +368,28 @@ impl Transcribed {
 mod tests {
   use super::*;

+   #[test]
+   fn test_transcribe_tiny_jfk_wav_whisper_rs_wav_channel() {
+      wav_channel(
+         Flags {
+            model_path: None,
+            model_buffer: Some(include_bytes!("../../models/ggml-tiny.en.bin").to_vec()),
+            threads: 8,
+            stream_step: 30,
+            stream_retain: 0.0,
+            stream_head: 0.0,
+            stream_tail: 0.0,
+            wav: Some("../gitea-whisper-rs/sys/whisper.cpp/bindings/go/samples/jfk.wav".to_string()),
+            debug: false,
+            stream_device: None,
+         },
+         move | result | {
+            assert!(result.is_ok());
+            assert_eq!(result.unwrap().to_string(), " And so my fellow Americans ask not what your country can do for you ask what you can do for your country.");
+         },
+      );
+   }
+
   #[test]
   fn test_transcribe_tiny_jfk_wav_whisper_rs() {
      wav(
@@ -374,16 +412,4 @@ mod tests {
         "../gitea-whisper-rs/sys/whisper.cpp/bindings/go/samples/jfk.wav".to_string(),
      );
   }
-
-   #[test]
-   fn test_transcribe_tiny_jfk_wav_candle() {
-      let wav_path = "../gitea-whisper-rs/sys/whisper.cpp/bindings/go/samples/jfk.wav".to_string();
-      let audio_data = f32_from_wav_file(&wav_path).unwrap();
-      let engine = new_engine_2().unwrap();
-      let result = engine.transcribe(&audio_data);
-      assert_eq!(
-         " And so my fellow Americans ask not what your country can do for you ask what you can do for your country.".to_string(),
-         result.unwrap().to_string(),
-      );
-   }
 }
--- a/todo.yaml
+++ b/todo.yaml
@@ -0,0 +1,12 @@
+todo:
+- wav to subtitles
+- compound words like checkmark vs check mark should destutter
+- whisper trims outside silence so head and tail never get hit
+- split on silence-ish instead of duration
+- rust-whisper warn when transcription time ~ input time
+scheduled: []
+done:
+- todo: need to overlap without ANY puctuation, which i can do by breaking into words
+  ts: Tue Jan  2 13:23:00 EST 2024
+- todo: overlap without stop words
+  ts: Wed Jan  3 03:22:14 EST 2024
--- a/wav_to_mkv.d/sc.jpg
+++ b/wav_to_mkv.d/sc.jpg
--- a/wav_to_mkv.d/wav_subtitles.rs
+++ b/wav_to_mkv.d/wav_subtitles.rs
@@ -0,0 +1,100 @@
+// This example is not going to build in this folder.
+// You need to copy this code into your project and add the dependencies whisper_rs and hound in your cargo.toml
+
+use hound;
+use std::fs::File;
+use std::io::Write;
+use whisper_rs::{FullParams, SamplingStrategy, WhisperContext};
+
+/// Loads a context and model, processes an audio file, and prints the resulting transcript to stdout.
+fn main() -> Result<(), &'static str> {
+   let args: Vec<String> = std::env::args().collect();
+
+    // Load a context and model.
+    let ctx = WhisperContext::new(&args[1])
+        .expect("failed to load model");
+    // Create a state
+    let mut state = ctx.create_state().expect("failed to create key");
+
+    // Create a params object for running the model.
+    // The number of past samples to consider defaults to 0.
+    let mut params = FullParams::new(SamplingStrategy::Greedy { best_of: 0 });
+
+    // Edit params as needed.
+    // Set the number of threads to use to 1.
+    //params.set_n_threads(1);
+    // Enable translation.
+    params.set_translate(true);
+    // Set the language to translate to to English.
+    params.set_language(Some("en"));
+    // Disable anything that prints to stdout.
+    params.set_print_special(false);
+    params.set_print_progress(false);
+    params.set_print_realtime(false);
+    params.set_print_timestamps(false);
+
+    // Open the audio file.
+    let mut reader = hound::WavReader::open(&args[2]).expect("failed to open file");
+    #[allow(unused_variables)]
+    let hound::WavSpec {
+        channels,
+        sample_rate,
+        bits_per_sample,
+        ..
+    } = reader.spec();
+
+    // Convert the audio to floating point samples.
+    let mut audio = whisper_rs::convert_integer_to_float_audio(
+        &reader
+            .samples::<i16>()
+            .map(|s| s.expect("invalid sample"))
+            .collect::<Vec<_>>(),
+    );
+
+    // Convert audio to 16KHz mono f32 samples, as required by the model.
+    // These utilities are provided for convenience, but can be replaced with custom conversion logic.
+    // SIMD variants of these functions are also available on nightly Rust (see the docs).
+    if channels == 2 {
+        audio = whisper_rs::convert_stereo_to_mono_audio(&audio)?;
+    } else if channels != 1 {
+        panic!(">2 channels unsupported");
+    }
+
+    if sample_rate != 16000 {
+        panic!("sample rate must be 16KHz");
+    }
+
+    // Run the model.
+    state.full(params, &audio[..]).expect("failed to run model");
+
+    // Create a file to write the transcript to.
+    let mut file = File::create("transcript.txt").expect("failed to create file");
+
+    // Iterate through the segments of the transcript.
+    let num_segments = state
+        .full_n_segments()
+        .expect("failed to get number of segments");
+    for i in 0..num_segments {
+        // Get the transcribed text and timestamps for the current segment.
+        let segment = state
+            .full_get_segment_text(i)
+            .expect("failed to get segment");
+        let start_timestamp = state
+            .full_get_segment_t0(i)
+            .expect("failed to get start timestamp");
+        let end_timestamp = state
+            .full_get_segment_t1(i)
+            .expect("failed to get end timestamp");
+
+        // Print the segment to stdout.
+        println!("[{} - {}]: {}", start_timestamp, end_timestamp, segment);
+
+        // Format the segment information as a string.
+        let line = format!("[{} - {}]: {}\n", start_timestamp, end_timestamp, segment);
+
+        // Write the segment information to the file.
+        file.write_all(line.as_bytes())
+            .expect("failed to write to file");
+    }
+    Ok(())
+}
--- a/wav_to_mkv.d/wav_to_mkv.sh
+++ b/wav_to_mkv.d/wav_to_mkv.sh
@@ -0,0 +1,66 @@
+#! /bin/bash
+
+main() {
+   set -euo pipefail
+
+   input_wav="$(realpath "$1")"
+   model="$(realpath "${2:-../models/ggml-small.en.bin}")"
+   already_transcribed="${3:-false}"
+
+   sanitized_wav="${input_wav%.*}.mono-16khz.wav"
+   ffmpeg -y -i "$input_wav" -ac 1 -ar 16k "$sanitized_wav"
+
+   if ! $already_transcribed; then
+      pushd "$(dirname "$(realpath "$BASH_SOURCE")")"
+      cd ../gitea-whisper-rs/
+      cargo run --example wav_subtitles -- "$model" "$sanitized_wav"
+      popd
+   fi
+   out_to_srt ../gitea-whisper-rs/transcript.txt > "${input_wav%.*}.srt"
+
+   ffmpeg -y \
+      -loop 1 -i sc.jpg \
+      -i "$input_wav" \
+      -i "${input_wav%.*}.srt" \
+      -c:v libx264 \
+      -tune stillimage \
+      -pix_fmt yuv420p -shortest \
+      "${input_wav%.*}.mkv"
+
+   ls "${input_wav%.*}.mkv"
+}
+
+out_to_srt() {
+   cs_to_ts() {
+      echo "$1" | awk '{
+         printf "%02d:%02d:%02d,000",
+            int(($1/100.0)/60/60),
+            int(($1/100.0)/60%60),
+            int(($1/100.0)%60)
+      }'
+   }
+
+   cat "$1" \
+   | (
+      i=0
+      while read -r line; do
+         ((i+=1))
+         echo "$i"
+         echo "$(cs_to_ts "$(
+            echo "${line%%:]  *}" \
+            | tr -d '[' \
+            | awk '{print $1}'
+         )") --> $(cs_to_ts "$(
+            echo "${line%%:]  *}" \
+            | tr -d '[' \
+            | awk '{print $3}'
+         )")"
+         echo "${line#*:  }"
+         echo
+      done
+   )
+}
+
+if [ "$0" == "$BASH_SOURCE" ]; then
+   main "$@"
+fi
--- a/1
+++ b/1
Author	SHA1	Message	Date
Bel LaPointe	fffea2ddf0	no render mac	2025-09-10 11:20:01 -06:00
Bel LaPointe	12dbf12299	k	2024-09-21 21:33:40 -04:00
Bel LaPointe	f04a55590f	fixed	2024-09-21 21:33:40 -04:00
Bel LaPointe	2254afcbfb	wav to mkv with subtitles scripting	2024-09-21 21:33:40 -04:00
bel	5fdc60e32c	stem words for destuttering	2024-01-03 20:40:35 -07:00
bel	4c80247ab9	accept lower sample rates if 16k not avail	2024-01-03 17:18:07 -07:00
bel	53e675b9a0	no panic on unusable mic	2024-01-03 17:09:27 -07:00
Bel LaPointe	9780c6f2ef	todo	2024-01-03 08:50:59 -07:00
Bel LaPointe	7f902af26f	default update	2024-01-03 08:40:13 -07:00
Bel LaPointe	9bc009996c	oop	2024-01-03 08:38:24 -07:00
Bel LaPointe	cbc8a4f9fd	cargo run -- --stream-step 8 --stream-retain 4 --stream-head=2 --stream-tail=0 2> /dev/null	2024-01-03 08:37:27 -07:00
Bel LaPointe	a8c8140d18	functionize at least	2024-01-03 08:28:22 -07:00
Bel LaPointe	5bc3209070	x=2; cargo run -- --wav $HOME/Downloads/41A6C472-6E4D-4953-9A90-2497D2DAD8C9.wav --stream-step $((x4)) --stream-retain $((x2)) --stream-{head,tail}=$((x)) 2> /dev/null	2024-01-03 08:22:45 -07:00
Bel LaPointe	8b5c18e65e	todo	2024-01-03 08:22:15 -07:00
Bel LaPointe	ec47d8142a	destutter with stopwords impl	2024-01-03 07:54:21 -07:00
bel	03659164ba	wip	2024-01-02 21:14:02 -07:00
bel	709dd1dba3	tod	2024-01-02 21:12:33 -07:00
bel	26595396cf	tod	2024-01-02 21:01:01 -07:00
Bel LaPointe	fb7892b52b	todo	2024-01-02 18:38:13 -07:00
Bel LaPointe	b08e055dac	todo	2024-01-02 18:23:03 -07:00
Bel LaPointe	9d993cfc8a	update destutterer to do punctuation-free words	2024-01-02 18:20:46 -07:00
Bel LaPointe	f4f8ea429a	merge	2024-01-02 17:51:29 -07:00
Bel LaPointe	38bea3735f	todo	2024-01-02 17:51:14 -07:00
Bel LaPointe	1c48026690	need to overlap without ANY puctuation, which i can do by breaking into words	2024-01-02 17:49:47 -07:00
Bel LaPointe	a57312786a	gr	2024-01-02 17:48:17 -07:00
Bel LaPointe	55e3bf0a26	update defaults	2024-01-02 17:47:00 -07:00
Bel LaPointe	743c8c5f67	time cargo run -- --wav $HOME/Downloads/41A6C472-6E4D-4953-9A90-2497D2DAD8C9.wav --stream-step 30 --stream-retain 25 --stream-{head,tail}=1 2> /dev/null	2024-01-02 16:45:04 -07:00
Bel LaPointe	d32f7a4c40	destutterer doesnt drop stutter for prev	2024-01-02 16:36:39 -07:00
Bel LaPointe	d94cbd6927	baked-lib wav_channel to base	2024-01-02 16:30:07 -07:00
Bel LaPointe	7a5db3b2ac	no callback on empty str	2024-01-02 16:29:58 -07:00
Bel LaPointe	d0dc9571d7	rust-whisper-baked listen also destutters	2024-01-02 16:12:07 -07:00
Bel LaPointe	6082f7e446	rust-whisper-baked de-stutters wav_channel output	2024-01-02 16:10:11 -07:00
Bel LaPointe	dd6f980266	rust-whisper-baked with --wav to wav_channel	2024-01-02 14:18:37 -07:00
Bel LaPointe	601fe517d7	rust-whisper-baked-lib::wav_channel()	2024-01-02 14:17:57 -07:00
Bel LaPointe	cd339de334	rust-whisper-lib::wav_channel()	2024-01-02 14:15:53 -07:00
bel	393100973c	baked streams even wav files	2024-01-01 20:56:53 -07:00
bel	97c025f04d	rust-whisper-baked works with WAV	2024-01-01 19:06:14 -07:00
bel	871efd9b8c	drop unused	2024-01-01 18:41:49 -07:00
Bel LaPointe	86bb1769f3	del unused	2023-12-21 22:30:43 -05:00