Add spectro webpage (not the audio of the server yet) & dataminer script

TODO

@@ -0,0 +1 @@
+- Fix install of venv

config/analyzer.conf

@@ -7,7 +7,7 @@ LOCATION="Maison ORTION - Saint-Gervais-en-Belin"
 # Species selection list
 SPECIES_LIST="./config/species_list.txt"
 # Minimal confidence threshold
-CONFIDENCE=0.25
+CONFIDENCE=0.1
 # Recording duration (in seconds)
 RECORDING_DURATION=15
 # Chunk folder location

daemon/birdnet_clean.sh

@@ -28,9 +28,9 @@ mem() {
     string=$2
     substring=$1
     if [[ "$string" == *"$substring"* ]]; then
-        echo "true"
+        true
     else
-        echo "false"
+        false
     fi
 }
 
@@ -49,7 +49,7 @@ junk() {
     # Get all empty treatment directories
     junk="$junk $(find ${CHUNK_FOLDER}/out -type d -empty)"
     # Get all empty record directories
-    treatement_folder=$(find "${CHUNK_FOLDER}/out" -type d ! -empty)
+    treatement_folder=$(find "${CHUNK_FOLDER}/out/*" -type d ! -empty)
     if [[ ! -z ${treatement_folder} ]]; then
         for folder in $treatement_folder; do
             echo $folder

daemon/birdnet_miner.sh

@@ -0,0 +1,120 @@
+#! /usr/bin/env bash
+# Extract observations from a model output folder
+#
+
+DEBUG=${DEBUG:-0}
+
+set -e
+
+debug() {
+    if [ $DEBUG -eq 1 ]; then
+        echo "$1"
+    fi
+}
+
+# Load bash library to deal with BirdNET-stream database
+source ./daemon/database/scripts/database.sh
+
+# Load config
+source ./config/analyzer.conf
+# Check config
+if [[ -z ${CHUNK_FOLDER} ]]; then
+    echo "CHUNK_FOLDER is not set"
+    exit 1
+else
+    if [[ ! -d ${CHUNK_FOLDER}/out ]]; then
+        echo "CHUNK_FOLDER does not exist: ${CHUNK_FOLDER}/out"
+        echo "Cannot extract observations."
+        exit 1
+    fi
+fi
+
+if [[ -z ${LATITUDE} ]]; then
+    echo "LATITUDE is not set"
+    exit 1
+fi
+
+if [[ -z ${LONGITUDE} ]]; then
+    echo "LONGITUDE is not set"
+    exit 1
+fi
+
+model_outputs() {
+    find -name "model.out.csv" -type f ! -empty
+}
+
+source_wav() {
+    model_output_path=$1
+    model_output_dir=$(dirname $model_output_path)
+    source_wav=$(basename $model_output_dir | rev | cut --complement -d"." -f1 | rev)
+    echo $source_wav
+}
+
+record_datetime() {
+    source_wav=$1
+    source_base=$(basename $source_wav .wav)
+    record_date=$(echo $source_base | cut -d"_" -f2)
+    record_time=$(echo $source_base | cut -d"_" -f3)
+    YYYY=$(echo $record_date | cut -c 1-4)
+    MM=$(echo $record_date | cut -c 5-6)
+    DD=$(echo $record_date | cut -c 7-8)
+    HH=$(echo $record_time | cut -c 1-2)
+    MM=$(echo $record_time | cut -c 3-4)
+    SS=$(echo $record_time | cut -c 5-6)
+    SSS="000"
+    date="$YYYY-$MM-$DD $HH:$MM:$SS.$SSS"
+    echo $date
+}
+
+save_observations() {
+    model_output_path=$1
+    source_audio=$(source_wav $model_output_path)
+    debug "Audio source: $source_audio"
+    observations=$(cat $model_output_path | tail -n +2)
+    IFS=$'\n'
+    for observation in $observations; do
+        if [[ -z "$observation" ]]; then
+            continue
+        fi
+        # debug "Observation: $observation"
+        start=$(echo "$observation" | cut -d"," -f1)
+        end=$(echo "$observation" | cut -d"," -f2)
+        scientific_name=$(echo "$observation" | cut -d"," -f3)
+        common_name=$(echo "$observation" | cut -d"," -f4)
+        confidence=$(echo "$observation" | cut -d"," -f5)
+        debug "Observation: $scientific_name ($common_name) from $start to $end with confidence $confidence"
+        taxon_id=$(get_taxon_id "$scientific_name")
+        if [[ -z $taxon_id ]]; then
+            debug "Taxon not found: $scientific_name"
+            debug "Inserting taxon..."
+            insert_taxon "$scientific_name" "$common_name"
+            taxon_id=$(get_taxon_id "$scientific_name")
+        fi
+        location_id=$(get_location_id "$LATITUDE" "$LONGITUDE")
+        if [[ -z $location_id ]]; then
+            debug "Location not found: $LATITUDE, $LONGITUDE"
+            debug "Inserting location..."
+            insert_location "$LATITUDE" "$LONGITUDE"
+            location_id=$(get_location_id "$LATITUDE" "$LONGITUDE")
+        fi
+        datetime=$(record_datetime $source_audio)
+        if [[ $(observation_exists "$source_audio" "$start" "$end" "$taxon_id" "$location_id") = "true" ]]; then
+            debug "Observation already exists: $source_audio, $start, $end, $taxon_id, $location_id"
+            exit 1
+        else
+            debug "Inserting observation: $source_audio, $start, $end, $taxon_id, $location_id"
+            insert_observation "$source_audio" "$start" "$end" "$taxon_id" "$location_id" "$confidence" "$datetime"
+        fi
+    done
+}
+
+main() {
+    # Remove all junk observations
+    ./daemon/birdnet_clean.sh
+    # Get model outputs
+    for model_output in $(model_outputs); do
+        save_observations $model_output
+    done
+}
+
+main

daemon/birdnet_recording.sh

@@ -32,7 +32,7 @@ record() {
     DEVICE=$1
     DURATION=$2
     debug "Recording from $DEVICE for $DURATION seconds"
-    echo "" | ffmpeg -nostdin -f pulse -i ${DEVICE} -t ${DURATION} -vn -acodec pcm_s16le -ac 1 -ar 48000 file:${CHUNK_FOLDER}/in/birdnet_$(date "+%Y%m%d_%H%M%S").wav
+    ffmpeg -nostdin -f pulse -i ${DEVICE} -t ${DURATION} -vn -acodec pcm_s16le -ac 1 -ar 48000 file:${CHUNK_FOLDER}/in/birdnet_$(date "+%Y%m%d_%H%M%S").wav
 }
 
 config_filepath="./config/analyzer.conf"

daemon/database/scripts/database.sh

@@ -0,0 +1,36 @@
+#! /usr/bin/env bash
+# SQLite library to deal with BirdNET-stream database
+
+set -e
+
+source ./config/analyzer.conf
+
+# Create database in case it was not created yet
+./daemon/database/scripts/create.sh
+
+DATABASE=${DATABASE:-"./var/db.sqlite"}
+
+get_location_id() {
+    sqlite3 $DATABASE "SELECT location_id FROM location WHERE latitude=$1 AND longitude=$2"
+}
+
+get_taxon_id() {
+    sqlite3 $DATABASE "SELECT taxon_id FROM taxon WHERE scientific_name='$1'"
+}
+
+insert_taxon() {
+    sqlite3 $DATABASE "INSERT INTO taxon (scientific_name, common_name) VALUES ('$1', '$2')"
+}
+
+insert_location() {
+    sqlite3 $DATABASE "INSERT INTO location (latitude, longitude) VALUES ($1, $2)"
+}
+
+insert_observation() {
+    sqlite3 $DATABASE "INSERT INTO observation (audio_file, start, end, taxon_id, location_id, confidence, date) VALUES ('$1', '$2', '$3', '$4', '$5', '$6', '$7')"
+}
+
+# Check if the observation already exists in the database
+observation_exists() {
+    sqlite3 $DATABASE "SELECT EXISTS(SELECT observation_id FROM observation WHERE audio_file='$1' AND start='$2' AND end='$3' AND taxon_id='$4' AND location_id='$5')"
+}

daemon/database/structure.sql

@@ -7,23 +7,24 @@ CREATE TABLE IF NOT EXISTS taxon (
     common_name TEXT NOT NULL
 );
 
-/** Locality table */
+/** Location table */
-CREATE TABLE IF NOT EXISTS locality (
+CREATE TABLE IF NOT EXISTS location (
-    locality_id INTEGER PRIMARY KEY,
+    location_id INTEGER PRIMARY KEY,
-    name TEXT NOT NULL,
     latitude REAL NOT NULL,
     longitude REAL NOT NULL
 );
 
 /** Observation table */
 CREATE TABLE IF NOT EXISTS observation (
-    observation_id INTEGER PRIMARY KEY,
+    `observation_id` INTEGER PRIMARY KEY,
-    taxon_id INTEGER NOT NULL,
+    `audio_file` TEXT NOT NULL,
-    locality_id INTEGER NOT NULL,
+    `start` REAL NOT NULL,
-    date TEXT NOT NULL,
+    `end` REAL NOT NULL,
-    time TEXT NOT NULL,
+    `taxon_id` INTEGER NOT NULL,
-    notes TEXT,
+    `location_id` INTEGER NOT NULL,
-    confidence REAL NOT NULL,
+    `date` TEXT NOT NULL,
+    `notes` TEXT,
+    `confidence` REAL NOT NULL,
     FOREIGN KEY(taxon_id) REFERENCES taxon(taxon_id),
-    FOREIGN KEY(locality_id) REFERENCES locality(locality_id)
+    FOREIGN KEY(location_id) REFERENCES location(location_id)
 );

install.sh

@@ -43,12 +43,12 @@ install_birdnetstream() {
     workdir=$(pwd)
     if [ -d "$workdir/BirdNET-stream" ]; then
         debug "BirdNET-stream is already installed"
-        return
+    else
-    fi
         # Clone BirdNET-stream
         debug "Cloning BirdNET-stream from $REPOSITORY"
         git clone --recurse-submodules $REPOSITORY
         # Install BirdNET-stream
+    fi
     cd BirdNET-stream
     debug "Creating python3 virtual environment '$PYTHON_VENV'"
     python3 -m venv $PYTHON_VENV

www/assets/app.js

@@ -10,3 +10,5 @@ import './styles/app.css';
 
 // start the Stimulus application
 import './bootstrap';
+
+import './utils/spectro';

www/assets/styles/app.css

@@ -81,3 +81,9 @@ nav ul li a:hover {
     box-shadow: 0px 8px 16px 0px rgba(0, 0, 0, 0.2);
     z-index: 1;
 }
+
+canvas {
+    display: block;
+    height: 100%;
+    width: 100%;
+}

www/assets/utils/spectro.js

@@ -0,0 +1,63 @@
+/**
+ * Credits to:
+ * https://codepen.io/jakealbaugh/pen/jvQweW
+ */
+
+// UPDATE: there is a problem in chrome with starting audio context
+//  before a user gesture. This fixes it.
+var started = false;
+var spectro_button = document.getElementById('spectro-button');
+spectro_button.addEventListener('click', () => {
+  if (started) return;
+  started = true;
+  console.log("starting spectro");
+  initialize();
+})
+
+function initialize() {
+  const CVS = document.getElementById('spectro-canvas');
+  const CTX = CVS.getContext('2d');
+  const W = CVS.width = window.innerWidth;
+  const H = CVS.height = window.innerHeight;
+
+  const ACTX = new AudioContext();
+  const ANALYSER = ACTX.createAnalyser();
+
+  ANALYSER.fftSize = 4096;  
+  
+  navigator.mediaDevices
+  .getUserMedia({ audio: true })
+  .then(process);
+
+  function process(stream) {
+    const SOURCE = ACTX.createMediaStreamSource(stream);
+    SOURCE.connect(ANALYSER);
+    const DATA = new Uint8Array(ANALYSER.frequencyBinCount);
+    const LEN = DATA.length;
+    const h = H / LEN;
+    const x = W - 1;
+    CTX.fillStyle = 'hsl(280, 100%, 10%)';
+    CTX.fillRect(0, 0, W, H);
+
+    loop();
+
+    function loop() {
+      window.requestAnimationFrame(loop);
+      let imgData = CTX.getImageData(1, 0, W - 1, H);
+      CTX.fillRect(0, 0, W, H);
+      CTX.putImageData(imgData, 0, 0);
+      ANALYSER.getByteFrequencyData(DATA);
+      for (let i = 0; i < LEN; i++) {
+        let rat = DATA[i] / 255;
+        let hue = Math.round((rat * 120) + 280 % 360);
+        let sat = '100%';
+        let lit = 10 + (70 * rat) + '%';
+        CTX.beginPath();
+        CTX.strokeStyle = `hsl(${hue}, ${sat}, ${lit})`;
+        CTX.moveTo(x, H - (i * h));
+        CTX.lineTo(x, H - (i * h + h));
+        CTX.stroke();
+      }
+    }
+  }
+}

www/nginx.conf

@@ -4,9 +4,8 @@ server {
 	root /var/www/html;
 
 	location / {
-		return 302 https://$host$request_uri;
+		return 302 https://$server_name$request_uri;
 	}
-	
 }
 
 server {

www/src/Controller/SpectroController.php

@@ -0,0 +1,21 @@
+<?php
+// src/Controller/AboutController.php
+namespace App\Controller;
+
+use Symfony\Component\HttpFoundation\Response;
+use Symfony\Component\HttpFoundation\Request;
+use Symfony\Component\Routing\Annotation\Route;
+use Symfony\Bundle\FrameworkBundle\Controller\AbstractController;
+
+class SpectroController extends AbstractController
+{
+    /**
+     * @Route("/spectro", name="spectro")
+     */
+    public function about()
+    {
+        return $this->render('spectro/index.html.twig', [
+        
+        ]);
+    }
+}

www/templates/spectro/index.html.twig

@@ -0,0 +1,9 @@
+{% extends "base.html.twig" %}
+
+{% block content %}
+    <h2>{{ "Spectrogram"  | trans }}</h2>
+    <div>
+        <button id="spectro-button">{{ "Launch Live Spectrogram" | trans }}</button>
+        <canvas id="spectro-canvas"></canvas>
+    </div>
+{% endblock %}