Add local spoken passages with Google retrieval and isolated runtime setup

apps/desktop/src-tauri/Cargo.lock

@@ -3809,6 +3809,7 @@ dependencies = [
  "searchi-core",
  "searchi-media",
  "searchi-service",
+ "searchi-speech",
  "serde",
  "serde_json",
  "sha2 0.10.9",
@@ -3859,11 +3860,22 @@ dependencies = [
  "searchi-core",
  "searchi-media",
  "searchi-semantic",
+ "searchi-speech",
  "serde",
  "sha2 0.10.9",
  "usearch",
 ]
 
+[[package]]
+name = "searchi-speech"
+version = "0.1.0"
+dependencies = [
+ "anyhow",
+ "serde",
+ "serde_json",
+ "sha2 0.10.9",
+]
+
 [[package]]
 name = "selectors"
 version = "0.38.0"

apps/desktop/src-tauri/Cargo.toml

@@ -26,6 +26,7 @@ sha2 = "0.10"
 tempfile = "3"
 searchi-core = { path = "../../../crates/searchi-core" }
 searchi-service = { path = "../../../crates/searchi-service" }
+searchi-speech = { path = "../../../crates/searchi-speech" }
 searchi-assets = { path = "../../../crates/searchi-assets" }
 searchi-media = { path = "../../../crates/searchi-media" }
 searchi-ai = { path = "../../../crates/searchi-ai" }

apps/desktop/src-tauri/src/main.rs

@@ -11,6 +11,7 @@ mod media_preview;
 mod model_download;
 mod semantic_commands;
 mod settings;
+mod speech_setup;
 mod watching;
 
 use ai_commands::{generate_ai, prepare_ai_source, AiSession};
@@ -650,7 +651,18 @@ fn main() {
                     "The media decoder is missing from this installation. Repair or reinstall searchi.".into()
                 })
             })();
+            let speech_worker = std::env::var_os("SEARCHI_SPEECH_WORKER").map(PathBuf::from).unwrap_or_else(|| {
+                            let backend = if std::env::var("SEARCHI_FORCE_CPU").as_deref() == Ok("1") { "cpu" } else { "auto" };
+                            let mut development = development_root.join(".searchi/speech-runtime").join(backend);
+                            if cfg!(windows) { development = development.join("Release"); }
+                            let name = if cfg!(windows) { "searchi-speech-worker.exe" } else { "searchi-speech-worker" };
+                            let development = development.join(name);
+                            if cfg!(debug_assertions) && development.is_file() { development } else {
+                                app.path().resource_dir().unwrap_or_else(|_| directory.clone()).join("speech-runtime").join(name)
+                            }
+                        });
             let mut ignored = vec![directory.clone(), model_directory.clone()];
+            if let Some(parent) = speech_worker.parent().filter(|path| path.is_absolute()) { ignored.push(parent.to_path_buf()); }
             if let Some(parent) = worker.parent().filter(|path| path.is_absolute()) {
                 ignored.push(parent.to_path_buf());
             }
@@ -667,7 +679,9 @@ fn main() {
             ).map_err(std::io::Error::other)?);
             let semantic = Arc::new(
                 SemanticState::new(model_directory, worker, media_tools)
-                    .with_media_scratch(directory.join("MediaScratch")),
+                    .with_media_scratch(directory.join("MediaScratch"))
+                    .with_speech_assets(
+speech_worker, directory.join("SpeechModels")),
             );
             semantic.begin_automatic_setup(database.clone(), settings.value.cpu_threads);
             #[cfg(target_os = "macos")]

apps/desktop/src-tauri/src/semantic_commands.rs

@@ -18,6 +18,7 @@ use tauri::State;
 pub struct SemanticState {
     pub model_directory: PathBuf,
     pub worker: PathBuf,
+    speech: std::sync::Arc<crate::speech_setup::SpeechSetup>,
     verified: AtomicBool,
     text_verified: AtomicBool,
     media_verified: AtomicBool,
@@ -48,7 +49,23 @@ impl SemanticState {
             Ok(tools) => (Some(tools), None),
             Err(error) => (None, Some(error)),
         };
+        let speech = crate::speech_setup::SpeechSetup::new(
+            worker
+                .parent()
+                .unwrap_or_else(|| std::path::Path::new("."))
+                .join("speech-runtime")
+                .join(if cfg!(windows) {
+                    "searchi-speech-worker.exe"
+                } else {
+                    "searchi-speech-worker"
+                }),
+            model_directory
+                .parent()
+                .unwrap_or_else(|| std::path::Path::new("."))
+                .join("speech"),
+        );
         Self {
+            speech,
             model_directory,
             worker,
             verified: AtomicBool::new(false),
@@ -72,6 +89,11 @@ impl SemanticState {
         }
     }
 
+    pub fn with_speech_assets(mut self, worker: PathBuf, models: PathBuf) -> Self {
+        self.speech = crate::speech_setup::SpeechSetup::new(worker, models);
+        self
+    }
+
     /// Keep generated video frames below the application's excluded data tree.
     pub fn with_media_scratch(mut self, directory: PathBuf) -> Self {
         self.media_limits.scratch_directory = Some(directory);
@@ -222,7 +244,12 @@ impl SemanticState {
             })
             .unwrap_or(Ok(()));
         self.warming.store(false, Ordering::Release);
-        self.finish_setup(provision, prepared)
+        let result = self.finish_setup(provision, prepared);
+        if result.is_ok() {
+            self.speech
+                .start(self.inference_budget.clone(), self.media_limits.clone());
+        }
+        result
     }
 
     fn finish_setup(
@@ -299,6 +326,7 @@ impl SemanticState {
 
 #[derive(Serialize)]
 pub struct SemanticStatus {
+    speech: crate::speech_setup::SpeechStatus,
     model: &'static str,
     ready: bool,
     text_model_ready: bool,
@@ -340,8 +368,11 @@ pub fn semantic_status(state: State<'_, AppState>) -> Result<SemanticStatus, Str
     let desired = semantic.desired_coverage();
     let index = HybridSearch::embedding_status_for_coverage(&state.database, desired)
         .map_err(|error| error.to_string())?;
-    let indexing_ready = semantic.can_index(index.text_indexing_allowed);
+    let speech = semantic.speech.status(&index);
+    let indexing_ready = semantic.can_index(index.text_indexing_allowed)
+        && (index.speech_policy.is_none() || speech.ready);
     Ok(SemanticStatus {
+        speech,
         model: "Google EmbeddingGemma 2",
         ready: worker_ready && semantic.warmed.load(Ordering::Acquire),
         text_model_ready: semantic.text_verified.load(Ordering::Acquire),
@@ -428,7 +459,9 @@ pub async fn index_embeddings(state: State<'_, AppState>) -> Result<EmbeddingRep
     let coverage = semantic.desired_coverage();
     let index = HybridSearch::embedding_status_for_coverage(&state.database, coverage)
         .map_err(|error| error.to_string())?;
-    if !semantic.can_index(index.text_indexing_allowed) {
+    if !semantic.can_index(index.text_indexing_allowed)
+        || (index.speech_policy.is_some() && !semantic.speech.status(&index).ready)
+    {
         return Err(
             "Semantic indexing is waiting for verified models and query preparation.".into(),
         );
@@ -464,6 +497,14 @@ pub async fn index_embeddings(state: State<'_, AppState>) -> Result<EmbeddingRep
                     .and_then(|report| report.ok_or_else(|| anyhow::anyhow!("Full media index retained while media setup finishes."))),
                 SemanticCoverage::FullMedia => {
                     let tools = semantic.media_tools.as_ref().ok_or_else(|| anyhow::anyhow!("The media decoder is missing from this installation. Repair or reinstall searchi."))?;
+                    let admission = semantic.inference_budget.acquire(InferenceRole::Background, 1)?;
+                    let speech = semantic.speech.worker_for_index()?;
+                    drop(admission);
+                    if let Some(speech) = speech {
+                        service = service.with_speech_provider(Box::new(searchi_service::BudgetedSpeechProvider::new(
+                            Box::new(speech), semantic.inference_budget.clone(),
+                        )))?;
+                    }
                     service.index_embeddings_with_media(tools, &semantic.media_limits)
                 }
             }

apps/desktop/src-tauri/src/speech_setup.rs

@@ -0,0 +1,120 @@
+//! Supplemental local speech setup; Google search readiness stays independent.
+use searchi_service::{EmbeddingStatus, InferenceBudget, InferenceRole, MediaLimits};
+use searchi_speech::{Device, SpeechWorker};
+use serde::Serialize;
+use std::{
+    path::PathBuf,
+    sync::{
+        atomic::{AtomicBool, AtomicU64, Ordering},
+        Arc, Mutex,
+    },
+};
+
+pub struct SpeechSetup {
+    worker: PathBuf,
+    models: PathBuf,
+    downloading: AtomicBool,
+    downloaded: AtomicU64,
+    ready: AtomicBool,
+    error: Mutex<Option<String>>,
+    policy: Mutex<Option<String>>,
+    prepared: Mutex<Option<SpeechWorker>>,
+}
+
+#[derive(Serialize)]
+pub struct SpeechStatus {
+    worker_ready: bool,
+    pub ready: bool,
+    downloading: bool,
+    downloaded_bytes: u64,
+    total_bytes: u64,
+    error: Option<String>,
+    requested_policy: Option<String>,
+    pub pending: bool,
+    corpus_revision: i64,
+    failures: usize,
+}
+
+impl SpeechSetup {
+    pub fn new(worker: PathBuf, models: PathBuf) -> Arc<Self> {
+        Arc::new(Self {
+            worker,
+            models,
+            downloading: AtomicBool::new(false),
+            downloaded: AtomicU64::new(0),
+            ready: AtomicBool::new(false),
+            error: Mutex::new(None),
+            policy: Mutex::new(None),
+            prepared: Mutex::new(None),
+        })
+    }
+
+    pub fn start(self: &Arc<Self>, budget: InferenceBudget, limits: MediaLimits) {
+        if self.downloading.swap(true, Ordering::AcqRel) {
+            return;
+        }
+        let setup = self.clone();
+        let spawned = std::thread::Builder::new().name("searchi-speech-setup".into()).spawn(move || {
+            let result = (|| -> anyhow::Result<()> {
+                anyhow::ensure!(setup.worker.is_file(), "Speech recognition is unavailable in this installation. Repair searchi to enable it.");
+                searchi_assets::speech::provision(&setup.models, &setup.downloaded)?;
+                let _admission = budget.acquire(InferenceRole::Background, 1)?;
+                let worker = setup.load()?;
+                let policy = searchi_service::speech_policy_key(worker.fingerprint(), &limits)?;
+                *setup.prepared.lock().map_err(|_| anyhow::anyhow!("Speech setup unavailable"))? = Some(worker);
+                *setup.policy.lock().map_err(|_| anyhow::anyhow!("Speech setup unavailable"))? = Some(policy);
+                Ok(())
+            })();
+            setup.ready.store(result.is_ok(), Ordering::Release);
+            if let Ok(mut error) = setup.error.lock() { *error = result.err().map(|error| error.to_string()); }
+            setup.downloading.store(false, Ordering::Release);
+        });
+        if let Err(error) = spawned {
+            if let Ok(mut last) = self.error.lock() {
+                *last = Some(format!("Could not start speech setup: {error}"));
+            }
+            self.downloading.store(false, Ordering::Release);
+        }
+    }
+
+    fn load(&self) -> anyhow::Result<SpeechWorker> {
+        SpeechWorker::load(
+            &self.worker,
+            &self.models.join("ggml-base.bin"),
+            &self.models.join("ggml-silero-v6.2.0.bin"),
+            Device::Auto,
+        )
+    }
+
+    pub fn worker_for_index(&self) -> anyhow::Result<Option<SpeechWorker>> {
+        if !self.ready.load(Ordering::Acquire) {
+            return Ok(None);
+        }
+        if let Some(worker) = self
+            .prepared
+            .lock()
+            .map_err(|_| anyhow::anyhow!("Speech setup unavailable"))?
+            .take()
+        {
+            return Ok(Some(worker));
+        }
+        self.load().map(Some)
+    }
+
+    pub fn status(&self, index: &EmbeddingStatus) -> SpeechStatus {
+        let policy = self.policy.lock().ok().and_then(|value| value.clone());
+        let ready = self.ready.load(Ordering::Acquire);
+        SpeechStatus {
+            worker_ready: self.worker.is_file(),
+            ready,
+            downloading: self.downloading.load(Ordering::Acquire),
+            downloaded_bytes: self.downloaded.load(Ordering::Acquire),
+            total_bytes: searchi_assets::speech::TOTAL_BYTES,
+            error: self.error.lock().ok().and_then(|value| value.clone()),
+            pending: ready && (index.speech_policy != policy || !index.speech_failures.is_empty()),
+            requested_policy: policy,
+            corpus_revision: index.corpus_revision,
+            failures: index.speech_failures.len(),
+        }
+    }
+}

apps/desktop/src/main.ts

@@ -104,7 +104,20 @@ interface DesktopStatus {
     watch_error: string | null;
   }[];
 }
+interface SpeechStatus {
+  worker_ready: boolean;
+  ready: boolean;
+  downloading: boolean;
+  downloaded_bytes: number;
+  total_bytes: number;
+  error: string | null;
+  requested_policy: string | null;
+  pending: boolean;
+  corpus_revision: number;
+  failures: number;
+}
 interface SemanticStatus {
+  speech?: SpeechStatus;
   model: string;
   ready: boolean;
   model_directory: string;
@@ -150,7 +163,7 @@ document.querySelector<HTMLDivElement>("#app")!.innerHTML = `
 <header><div class="brand"><img class="brand-symbol" src="/searchi-symbol.png" alt=""><span class="wordmark">searchi</span></div><span id="index-status" class="index-status">Loading your index…</span><div class="header-actions"><button id="add-folder" class="quiet-button" type="button">${icon("folderAdd")}<span>Add folder</span></button><details id="filters" class="popover"><summary>${icon("filter")}<span id="scope-label">All files</span>${icon("chevron")}</summary><div class="popover-panel"><h2>Filter your search</h2><div class="filters-grid"><label>Search scope<select id="filter-scope"><option value="all">All files</option><option value="code">Code</option></select></label><label>Within folder<input id="filter-folder" placeholder="Absolute folder path"></label><label>Extension<input id="filter-extension" placeholder="pdf, md, txt…"></label><label>Artist (exact)<input id="filter-artist" placeholder="Artist from music tags" maxlength="256"></label><label>Track title (exact)<input id="filter-title" placeholder="Title from music tags" maxlength="256"></label><label>Minimum size (bytes)<input id="filter-min" type="number" min="0" step="1"></label><label>Maximum size (bytes)<input id="filter-max" type="number" min="0" step="1"></label><label>Minimum duration (seconds)<input id="filter-duration-min" type="number" min="0" step="0.1"></label><label>Maximum duration (seconds)<input id="filter-duration-max" type="number" min="0" step="0.1"></label><label>Modified from<input id="filter-after" type="date"></label><label>Modified through<input id="filter-before" type="date"></label></div></div></details><details id="settings" class="popover"><summary class="icon-button" aria-label="Settings" title="Settings">${icon("settings")}</summary><div class="popover-panel"><h2>Settings</h2><section class="appearance-settings" aria-labelledby="appearance-heading"><h3 id="appearance-heading">Appearance</h3><label for="background-material">Window background</label><select id="background-material"><option value="solid">Solid</option><option value="glass">Glass</option><option value="frosted">Frosted glass</option></select><div class="appearance-opacity"><label for="background-opacity">Background tint</label><output id="background-opacity-value" for="background-opacity">90%</output></div><input id="background-opacity" type="range" min="70" max="100" step="1" value="90" aria-describedby="appearance-status"><p id="appearance-status" class="hint" role="status">A subtle window background. Search results and passages stay clear.</p></section><h3>Folders & resources</h3><p id="shortcut-status" class="hint"></p><label class="resource-label" for="cpu-threads">CPU thread budget</label><select id="cpu-threads">${Array.from({ length: 8 }, (_, i) => `<option value="${i + 1}">${i + 1} ${i === 1 ? "threads (default)" : i ? "threads" : "thread (low power)"}</option>`).join("")}</select><p class="hint">Search automatically uses supported acceleration with CPU fallback. Two or more threads let search run alongside indexing; one thread can delay semantic search during indexing.</p><ul id="roots" aria-label="Selected folders"></ul><label class="performance-toggle"><input id="performance-toggle" type="checkbox">Performance details</label><p class="hint">Local timing diagnostics for new searches. Queries and file paths are excluded from copied reports.</p><details id="ai-settings" class="ai-settings"><summary>Optional AI answers</summary><label class="performance-toggle"><input id="ai-enabled" type="checkbox">Enable summaries and questions</label><p class="hint">Disabled by default. Only an explicit action sends the shown passage to your configured endpoint.</p><label>OpenAI-compatible base URL<input id="ai-endpoint" type="url" placeholder="http://localhost:1234/v1" autocomplete="off" maxlength="2048"></label><label>Model<input id="ai-model" placeholder="Your model name" autocomplete="off" maxlength="200"></label><label>API key (optional, this session only)<input id="ai-key" type="password" autocomplete="off" maxlength="2048"></label><button id="ai-clear-key" type="button">Clear key</button><p class="hint">Endpoint, model and key stay in this app session. The key authenticates requests to your endpoint.</p></details></div></details></div></header>
 <main>
 <form id="search-form" role="search"><label class="sr-only" for="query">Search your files</label>${icon("search")}<input id="query" type="search" placeholder="Search a filename, phrase, or what you remember…" autocomplete="off" spellcheck="false" maxlength="1000"><kbd id="focus-shortcut">Ctrl K</kbd><button id="clear-query" class="icon-button" type="button" aria-label="Clear search" title="Clear search">${icon("dismiss")}</button></form>
-<details id="semantic-panel"><summary>${icon("sparkles")}<span id="semantic-status" role="status" aria-live="polite">Preparing local semantic search…</span><span id="semantic-device" class="kind">Local model</span>${icon("chevron")}</summary><div class="semantic-content"><strong>Google EmbeddingGemma 2</strong><progress id="semantic-progress" aria-label="Local model setup progress"></progress><p id="semantic-detail" class="hint">The verified local model downloads automatically on first launch (865 MB). Your files stay on this device.</p><div class="semantic-actions"><button id="retry-model" type="button" hidden>Retry model setup</button><button id="rebuild-semantic" type="button" hidden>Rebuild semantic index</button></div></div></details>
+<details id="semantic-panel"><summary>${icon("sparkles")}<span id="semantic-status" role="status" aria-live="polite">Preparing local semantic search…</span><span id="semantic-device" class="kind">Local model</span>${icon("chevron")}</summary><div class="semantic-content"><strong>Google EmbeddingGemma 2</strong><progress id="semantic-progress" aria-label="Local model setup progress"></progress><p id="semantic-detail" class="hint">The verified local model downloads automatically on first launch (865 MB). Your files stay on this device.</p><p id="speech-status" class="hint" role="status" hidden></p><progress id="speech-progress" aria-label="Local speech model setup progress" hidden></progress><div class="semantic-actions"><button id="retry-model" type="button" hidden>Retry model setup</button><button id="rebuild-semantic" type="button" hidden>Rebuild semantic index</button></div></div></details>
 <form id="folder-form" hidden><label for="folder">Folder to index</label><div class="folder-row"><input id="folder" placeholder="Absolute folder path" required><button id="browse-folder" type="button">Browse…</button><button type="submit">Index folder</button><button id="cancel-folder" type="button">Cancel</button></div><p class="hint">Only choose folders you want searched. Large folders may take a while.</p></form>
 <p id="notice" role="status" aria-live="polite"></p><details id="performance-panel" hidden><summary>Performance details</summary><div id="performance-detail"></div><button id="copy-performance" type="button">Copy local timing report</button><span id="copy-performance-status" role="status"></span></details><div class="workspace"><section class="results-pane" aria-label="Search results"><p id="result-summary" class="eyebrow">READY WHEN YOU ARE</p><ul id="results" aria-label="Matching files"></ul><div id="empty">${icon("search")}<h2>Find what matters. Locally.</h2><p>Index a folder, then search a filename, phrase, or what you remember.</p></div></section><aside id="preview" aria-label="Result preview" hidden><div class="preview-header"><span id="preview-icon" class="file-icon"></span><div class="preview-heading"><h2 id="preview-title"></h2><p id="preview-path" class="path"></p></div></div><div class="preview-toolbar"><span id="preview-kind"></span><div class="preview-actions"><button id="open-result" type="button">${icon("open")}<span>Open file</span></button><button id="reveal-result" type="button" title="Show in folder" aria-label="Show in folder">${icon("folder")}</button></div></div><details id="match-details" class="match-details"><summary>Why this result?</summary><p id="match-reason"></p><p id="match-legend"></p></details><div id="media-preview" class="media-preview" hidden></div><p id="media-preview-status" class="hint" role="status" hidden></p><details id="ai-panel" class="ai-panel" hidden><summary>Summarize or ask</summary><p id="ai-disclosure" class="hint"></p><p class="eyebrow">PASSAGE THAT WILL BE SENT</p><pre id="ai-source-text"></pre><div class="ai-actions"><button id="ai-summarize" type="button" disabled>Summarize this passage</button><label class="sr-only" for="ai-question">Question about this passage</label><input id="ai-question" placeholder="Ask about this passage…" maxlength="2048"><button id="ai-ask" type="button" disabled>Ask</button></div><p id="ai-status" class="hint" role="status"></p><pre id="ai-answer"></pre><p id="ai-answer-label" class="hint"></p><button id="ai-citation" type="button" hidden>Selected source</button></details><div class="passage"><p id="passage-label" class="eyebrow">EXTRACTED PASSAGE</p><pre id="preview-text"></pre></div></aside></div>
 </main><footer><span class="footer-privacy">${icon("lock")}<span id="privacy-summary">Private search. Everything stays on your device.</span><span id="privacy" class="privacy">Local</span></span><span class="keyboard-hints"><kbd>${icon("arrowUp")}</kbd><kbd>${icon("arrowDown")}</kbd><span>Navigate</span><kbd>Enter</kbd><span>Open</span><kbd>Esc</kbd><span>Clear</span></span></footer>`;
@@ -452,9 +465,31 @@ async function refreshSemanticStatus() {
       element("semantic-detail").textContent +=
         ` ${status.media_failures.length} files could not be indexed. The first issue: ${status.media_failures[0].path}: ${status.media_failures[0].message}. Filename search remains available; rebuild after fixing the file.`;
     }
+    const speech = status.speech;
+    const speechStatus = element("speech-status");
+    speechStatus.hidden = !speech;
+    const speechProgress = element<HTMLProgressElement>("speech-progress");
+    speechProgress.hidden = !speech?.downloading;
+    if (speech) {
+      speechProgress.max = speech.total_bytes;
+      speechProgress.value = speech.downloaded_bytes;
+      speechStatus.textContent = speech.downloading
+        ? `Preparing local speech passages · ${Math.round(speech.downloaded_bytes / 1_000_000)} of ${Math.round(speech.total_bytes / 1_000_000)} MB`
+        : speech.error
+          ? `Speech passages: ${speech.error}`
+          : speech.ready
+            ? speech.failures
+              ? `Speech passages incomplete for ${speech.failures} files. Other search results remain available.`
+              : speech.pending
+                ? "Speech models ready · Spoken passages are waiting to be indexed."
+                : "Local speech passages ready. Recognized words and timings can contain errors."
+            : "Local speech passages are preparing. Filename and existing content search remain available.";
+    }
     const retry = element<HTMLButtonElement>("retry-model");
-    retry.hidden = !semanticRetryVisible({ ...status, error });
-    retry.disabled = working || isCaptureSurface();
+    retry.hidden =
+      !semanticRetryVisible({ ...status, error }) && !speech?.error;
+    retry.disabled =
+      working || Boolean(speech?.downloading) || isCaptureSurface();
     const rebuild = element<HTMLButtonElement>("rebuild-semantic");
     rebuild.hidden = !status.ready || (!status.chunks && !status.pending);
     rebuild.disabled =
@@ -467,8 +502,9 @@ async function refreshSemanticStatus() {
     if (
       !isCaptureSurface() &&
       status.ready &&
-      status.stale &&
-      status.pending > 0 &&
+      ((status.stale && status.pending > 0) ||
+        (Boolean(speech?.pending) &&
+          Boolean(status.generation !== null || status.pending))) &&
       semanticIndexReady(status) &&
       !semanticJobActive &&
       !indexingActive

apps/desktop/src/match-display.ts

@@ -224,6 +224,12 @@ export function matchExplanation(result: MatchResult) {
         legend: literalLegend,
         semanticPassage: false,
       };
+    case "semantic_speech":
+      return {
+        reason: `Related spoken passage${time}. Recognized locally; its time is an estimated token position.`,
+        legend: `${literalLegend} Speech recognition may contain mistakes.`,
+        semanticPassage: true,
+      };
     case "hybrid":
       return {
         reason: `Query words and related meaning matched this file; showing the passage${location}.`,

apps/desktop/src/media-display.ts

@@ -34,6 +34,7 @@ export function matchLabel(kind: string): string {
     semantic_image: "Image meaning",
     semantic_video: "Video frame meaning",
     semantic_audio: "Audio meaning",
+    semantic_speech: "Spoken passage",
     semantic: "Meaning",
     hybrid: "Keyword + meaning",
     exact_filename: "Exact filename",

apps/desktop/src/semantic-setup-state.ts

@@ -14,6 +14,12 @@ export interface SemanticSetupStatus {
   media_model_ready?: boolean;
   media_model_error?: string | null;
   requested_coverage?: SemanticCoverage;
+  speech?: {
+    ready: boolean;
+    pending: boolean;
+    requested_policy: string | null;
+    corpus_revision: number;
+  };
 }
 
 export function semanticIndexReady(status: SemanticSetupStatus) {
@@ -30,6 +36,11 @@ export function semanticIndexKey(
   status: SemanticSetupStatus,
   folderRevision: number,
 ) {
+  if (status.speech?.ready && status.speech.pending) {
+    // Failed optional speech can still publish raw media: publication must not
+    // create an infinite retry loop for the same source corpus and policy.
+    return `speech:${status.speech.requested_policy}:${status.speech.corpus_revision}:${folderRevision}`;
+  }
   return `${status.generation}:${status.pending}:${folderRevision}:${status.requested_coverage ?? "full_media"}`;
 }
 

apps/desktop/tests/match-display.test.mjs

@@ -74,6 +74,9 @@ test('optional reasons use real source routes and never claim semantic word conf
   assert.match(text.legend,/without the same words/);
   assert.equal(text.semanticPassage,true);
   assert.match(matchExplanation({match_kind:'semantic_audio',source:{kind:'timestamp_ms',number:13129}}).reason,/0:13.*not an exact spoken phrase/);
+  const speech=matchExplanation({match_kind:'semantic_speech',source:{kind:'timestamp_ms',number:13129}});
+  assert.match(speech.reason,/0:13.*estimated token position/);
+  assert.equal(speech.semanticPassage,true);
   assert.match(matchExplanation({match_kind:'semantic_video',source:{kind:'timestamp_ms',number:24167}}).reason,/frame.*0:24/);
   assert.match(matchExplanation({match_kind:'semantic_image'}).reason,/whole/);
   assert.match(matchExplanation({match_kind:'name_terms'}).reason,/filename, path or content/);

apps/desktop/tests/semantic-setup-state.test.mjs

@@ -51,3 +51,13 @@ test('legacy full-pair status remains blocked during download and ready after se
   assert.equal(semanticIndexReady({...legacy,downloading:false}), true);
   assert.equal(semanticMediaReady({...legacy,downloading:false}), true);
 });
+
+
+test('partial speech publication cannot create an automatic retry loop', () => {
+  const gate = new AutomaticIndexGate();
+  const first = {...text(), speech:{ready:true,pending:true,requested_policy:'speech-v1-a',corpus_revision:7}};
+  assert.equal(gate.claim(semanticIndexKey(first,4)),true);
+  assert.equal(gate.claim(semanticIndexKey({...first,generation:99,pending:0},4)),false);
+  assert.equal(gate.claim(semanticIndexKey({...first,speech:{...first.speech,corpus_revision:8}},4)),true);
+  assert.equal(gate.claim(semanticIndexKey({...first,speech:{...first.speech,requested_policy:'speech-v1-b'}},4)),true);
+});