wip

This reverts commit a483aaf25c.

rust-whisper.d/Cargo.lock

@@ -261,20 +261,21 @@ dependencies = [
 
 [[package]]
 name = "whisper-rs"
-version = "0.5.0"
+version = "0.8.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fa7e1b9b003aa3285a0e4469219566266aa1d51ced1be38587251a4f713a1677"
+checksum = "2c950fb18ad556b053ba615b88fd4d01ed6020be740c3371eb0fc4aec64a0639"
 dependencies = [
  "whisper-rs-sys",
 ]
 
 [[package]]
 name = "whisper-rs-sys"
-version = "0.3.1"
+version = "0.6.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "97a389dc665c7354ba6b1982850d4ba05b862907e535708ebdec92cbd9c599e8"
+checksum = "094a5bd86f6f52562bbc74c28f27cd80197e54656cfb7213cf4ba37b5246cc9e"
 dependencies = [
  "bindgen",
+ "cfg-if",
 ]
 
 [[package]]

rust-whisper.d/Cargo.toml

@@ -6,6 +6,6 @@ edition = "2021"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
-whisper-rs = "0.5"
+whisper-rs = "0.8.0"
 wav = "1"
 tokio = "1.27"

rust-whisper.d/hotwords.py

@@ -47,7 +47,7 @@ class Reader(threading.Thread):
       self.name = os.environ.get("MIC_NAME", "pulse_monitor")
       if not self.name:
          for index, name in enumerate(sr.Microphone.list_microphone_names()):
-             print("[{0}] Microphone with name \"{1}\" found for `Microphone(device_index={0})`".format(index, name))
+             log("[{0}] Microphone with name \"{1}\" found for `Microphone(device_index={0})`".format(index, name))
          exit()
       self.inq = inq
       self.outq = outq
@@ -136,14 +136,19 @@ class Parser(threading.Thread):
       p = "/tmp/whisper-cpp.wav"
       with open("/tmp/whisper-cpp.wav", "wb") as f:
          f.write(wav)
-      proc = subprocess.run(f"MODEL=./models/ggml-{os.environ.get('MODEL','tiny.en')}.bin WAV={p} P=2 rust-whisper", capture_output=True, shell=True)
+      proc = subprocess.run(f"MODEL=./models/ggml-{os.environ.get('MODEL','tiny.en')}.bin WAV={p} P={os.environ.get('P', '2')} rust-whisper", capture_output=True, shell=True)
       result = proc.stdout.decode().strip()
       if os.environ.get("DEBUG", None):
          log("stderr:", proc.stderr.decode().strip())
          log("raw transcript:", result)
       result = result.replace(">>", "")
-      result = "".join([i.split("]")[-1] for i in result.split("[")[0]])
+      for pair in [
-      result = "".join([i.split(")")[-1] for i in result.split("(")[0]])
+         ("[", "]"),
+         ("(", ")"),
+         ("<", ">"),
+         ("*", "*"),
+      ]:
+         result = "".join([i.split(pair[1])[-1] for i in result.split(pair[0])[0]])
       if os.environ.get("DEBUG", None):
          log("annotation-free transcript:", result)
       return result
@@ -180,6 +185,8 @@ def _load_dot_notation(v, items):
                else:
                   result.append(subresult)
             return result
+      elif k == "KEYS":
+         v = [k for k in v]
       else:
          if isinstance(v, list):
             v = v[int(k)]
@@ -224,8 +231,15 @@ class Reactor(threading.Thread):
             def load_hotwords_in_yaml_file():
                with open(p.split("@")[0], "r") as f:
                   v = yaml.safe_load(f)
-                  v = load_dot_notation(v, p.split("@")[-1])
+                  if os.environ.get("DEBUG", None):
-                  return ["".join(i.strip().lower().split()) for i in v if i]
+                     log(f'opened {p.split("@")[0]} and got {v}')
+                  result = []
+                  for to_find in [i for i in p.split("@")[-1].split(",") if i]:
+                     if os.environ.get("DEBUG", None):
+                        log(f'finding {to_find} in {v}')
+                     v2 = load_dot_notation(v, to_find)
+                     result.extend(["".join(i.strip().lower().split()) for i in v2 if i])
+                  return result
             load_hotwords_in_yaml_file()
             return load_hotwords_in_yaml_file
          else:
@@ -254,13 +268,23 @@ class Reactor(threading.Thread):
       log("Reactor.run: stop")
 
    def handle(self, text):
+      try:
+         self._handle(text)
+      except Exception:
+         pass
+
+   def _handle(self, text):
       hotwords = self.load_hotwords()
       if os.environ.get("DEBUG", None):
-         log(f"seeking {hotwords} in {text}")
+         log(f"seeking {hotwords} in {text}. $HOTWORDS={os.environ.get('HOTWORDS', None)}")
       if not hotwords:
          if not os.environ.get("HOTWORDS", None):
-            print(text)
+            if os.environ.get("DEBUG", None):
+               log(f"HOTWORDS is False; {text}")
+            print(text, flush=True)
          else:
+            if os.environ.get("DEBUG", None):
+               log(f"HOTWORDS is True; {text}")
             log(text)
          return
       cleantext = "".join([i for i in "".join(text.lower().split()) if i.isalpha()])
@@ -300,7 +324,7 @@ class Actor(threading.Thread):
 
    def handle_stdout(self, hotword, context):
       log(context)
-      print(hotword)
+      print(hotword, flush=True)
 
    def handle_signal(self, hotword, context):
       self.handle_stderr(hotword, context)
@@ -319,9 +343,12 @@ class Actor(threading.Thread):
          body = self.body
          body = body.replace("{{hotword}}", hotword)
          body = body.replace("{{context}}", context)
+         url = self.url
+         url = url.replace("{{hotword}}", hotword)
+         url = url.replace("{{context}}", context)
          if os.environ.get("DEBUG", "") :
-            log("POST", self.url, headers, body)
+            log("POST", url, headers, body)
-         requests.post(self.url, headers=headers, data=body)
+         requests.post(url, headers=headers, data=body)
       except Exception as e:
          log("Actor.handle_url:", e)
 

rust-whisper.d/models/download_models.sh

@@ -1,10 +1,9 @@
 #!/bin/bash
 
 src="https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml"
-d="${1:-"$PWD"/models}"
+cd "$(dirname "$(realpath "$BASH_SOURCE")")"
-mkdir -p "$d"
 
 # Whisper models
 for model in "tiny.en" "base.en" "small.en" "medium.en"; do
-    	test -f "$d"/ggml-$model.bin || wget --quiet --show-progress -O "$d"/ggml-$model.bin "$src-$model.bin"
+    	test -f ./ggml-$model.bin || wget --quiet --show-progress -O ./ggml-$model.bin "$src-$model.bin"
 done

rust-whisper.d/src/main.rs

@@ -1,9 +1,10 @@
 use whisper_rs::{WhisperContext, FullParams, SamplingStrategy};
 
 fn main() {
-   let mut ctx = WhisperContext::new(
+   let ctx = WhisperContext::new(
       &std::env::var("MODEL").unwrap_or(String::from("../models/ggml-tiny.en.bin"))
    ).expect("failed to load model");
+   let mut state = ctx.create_state().expect("failed to create state");
 
    // create a params object
    let mut params = FullParams::new(SamplingStrategy::Greedy { best_of: 0 });
@@ -25,12 +26,11 @@ fn main() {
    let data16 = data.as_sixteen().expect("wav is not 32bit floats");
    let audio_data = &whisper_rs::convert_integer_to_float_audio(&data16);
 
-   ctx.full(params, &audio_data[..])
+   state.full(params, &audio_data[..]).expect("failed to run model");
-      .expect("failed to run model");
 
-   let num_segments = ctx.full_n_segments();
+   let num_segments = state.full_n_segments().expect("failed to get number of segments");
    for i in 0..num_segments {
-      let segment = ctx.full_get_segment_text(i).expect("failed to get segment");
+      let segment = state.full_get_segment_text(i).expect("failed to get segment");
       print!("{} ", segment);
    }
    println!("");

rust-whisper.d/transcript.sh

@@ -0,0 +1,2 @@
+echo "pkill -9 -f hotwords.py; MIC_TIMEOUT=30 MODEL=small.en P=4 DEBUG=true HOTWORDS= python3 ./hotwords.py | tee -a $HOME/Sync/drawful/DnD/bdoob/__log.d/$(date +%Y.%m.%d).transcript.txt"
+echo "pkill -9 -f hotwords.py; MIC_TIMEOUT=30 MODEL=small.en P=4 DEBUG=true HOTWORDS= python3 ./hotwords.py | tee -a $HOME/Sync/drawful/DnD/nessira.d/_log.d/$(date +%Y.%m.%d).transcript.txt"