From 1be72ab5ece657fac42a025a73cf0dddc32b3854 Mon Sep 17 00:00:00 2001
From: neilruaro-camb <neil.ruaro@camb.ai>
Date: Thu, 9 Apr 2026 12:47:59 +0800
Subject: [PATCH] feat: add CAMB AI text-to-speech for note summaries

---
 .../pages/recording/RecordingDesktop.tsx      | 11 ++-
 .../pages/recording/RecordingMobile.tsx       |  8 ++-
 convex/camb.ts                                | 69 +++++++++++++++++++
 convex/schema.ts                              |  2 +
 convex/together.ts                            |  8 +++
 5 files changed, 96 insertions(+), 2 deletions(-)
 create mode 100644 convex/camb.ts

diff --git a/components/pages/recording/RecordingDesktop.tsx b/components/pages/recording/RecordingDesktop.tsx
index 7742ac8..af16808 100644
--- a/components/pages/recording/RecordingDesktop.tsx
+++ b/components/pages/recording/RecordingDesktop.tsx
@@ -20,6 +20,7 @@ export default function RecordingDesktop({
     transcription,
     title,
     _creationTime,
+    summaryAudioUrl,
   } = note;
   const [originalIsOpen, setOriginalIsOpen] = useState<boolean>(true);
 
@@ -85,7 +86,15 @@ export default function RecordingDesktop({
       <div className="grid h-full w-full grid-cols-2 px-[30px] lg:px-[45px]">
         <div className="relative min-h-[70vh] w-full border-r px-5 py-3 text-justify text-xl font-[300] leading-[114.3%] tracking-[-0.6px] lg:text-2xl">
           {transcription ? (
-            <div className="">{originalIsOpen ? transcription : summary}</div>
+            <div>
+              <div className="">{originalIsOpen ? transcription : summary}</div>
+              {!originalIsOpen && summaryAudioUrl && (
+                <div className="mt-6 flex items-center gap-3">
+                  <p className="text-sm opacity-60">Listen to summary</p>
+                  <audio controls src={summaryAudioUrl} className="h-8" />
+                </div>
+              )}
+            </div>
           ) : (
             // Loading state for transcript
             <ul className="animate-pulse space-y-3">
diff --git a/components/pages/recording/RecordingMobile.tsx b/components/pages/recording/RecordingMobile.tsx
index e2f0c60..1100753 100644
--- a/components/pages/recording/RecordingMobile.tsx
+++ b/components/pages/recording/RecordingMobile.tsx
@@ -12,7 +12,7 @@ export default function RecordingMobile({
   note: Doc<'notes'>;
   actionItems: Doc<'actionItems'>[];
 }) {
-  const { summary, transcription, title, _creationTime } = note;
+  const { summary, transcription, title, _creationTime, summaryAudioUrl } = note;
   const [transcriptOpen, setTranscriptOpen] = useState<boolean>(true);
   const [summaryOpen, setSummaryOpen] = useState<boolean>(false);
   const [actionItemOpen, setActionItemOpen] = useState<boolean>(false);
@@ -78,6 +78,12 @@ export default function RecordingMobile({
         {summaryOpen && (
           <div className="relative mt-2 min-h-[70vh] w-full px-4 py-3 text-justify font-light">
             {summary}
+            {summaryAudioUrl && (
+              <div className="mt-4 flex items-center gap-3">
+                <p className="text-sm opacity-60">Listen to summary</p>
+                <audio controls src={summaryAudioUrl} className="h-8" />
+              </div>
+            )}
           </div>
         )}
         {actionItemOpen && (
diff --git a/convex/camb.ts b/convex/camb.ts
new file mode 100644
index 0000000..522df3f
--- /dev/null
+++ b/convex/camb.ts
@@ -0,0 +1,69 @@
+("use node");
+
+import { internalAction, internalMutation } from "./_generated/server";
+import { v } from "convex/values";
+import { internal } from "./_generated/api";
+
+const CAMB_API_BASE = "https://client.camb.ai/apis";
+
+function getCambApiKey(): string {
+  const key = process.env.CAMB_API_KEY;
+  if (!key) {
+    throw new Error("CAMB_API_KEY environment variable is not set");
+  }
+  return key;
+}
+
+// Generate TTS audio of a note summary using CAMB AI
+export const generateSpeech = internalAction({
+  args: {
+    id: v.id("notes"),
+    text: v.string(),
+  },
+  handler: async (ctx, args) => {
+    const apiKey = getCambApiKey();
+
+    const res = await fetch(`${CAMB_API_BASE}/tts-stream`, {
+      method: "POST",
+      headers: {
+        "x-api-key": apiKey,
+        "Content-Type": "application/json",
+      },
+      body: JSON.stringify({
+        text: args.text,
+        voice_id: 147320,
+        language: "en-us",
+        speech_model: "mars-flash",
+        output_configuration: { format: "wav" },
+      }),
+    });
+
+    if (!res.ok) {
+      throw new Error(`CAMB TTS failed: ${await res.text()}`);
+    }
+
+    const audioBuffer = await res.arrayBuffer();
+    const blob = new Blob([audioBuffer], { type: "audio/wav" });
+
+    const storageId = await ctx.storage.store(blob);
+
+    await ctx.runMutation(internal.camb.saveSummaryAudio, {
+      id: args.id,
+      summaryAudioFileId: storageId,
+    });
+  },
+});
+
+export const saveSummaryAudio = internalMutation({
+  args: {
+    id: v.id("notes"),
+    summaryAudioFileId: v.id("_storage"),
+  },
+  handler: async (ctx, args) => {
+    const url = await ctx.storage.getUrl(args.summaryAudioFileId);
+    await ctx.db.patch(args.id, {
+      summaryAudioFileId: args.summaryAudioFileId,
+      summaryAudioUrl: url ?? undefined,
+    });
+  },
+});
diff --git a/convex/schema.ts b/convex/schema.ts
index 9e29a9b..1438da6 100644
--- a/convex/schema.ts
+++ b/convex/schema.ts
@@ -10,6 +10,8 @@ export default defineSchema({
     transcription: v.optional(v.string()),
     summary: v.optional(v.string()),
     embedding: v.optional(v.array(v.float64())),
+    summaryAudioFileId: v.optional(v.id('_storage')),
+    summaryAudioUrl: v.optional(v.string()),
     generatingTranscript: v.boolean(),
     generatingTitle: v.boolean(),
     generatingActionItems: v.boolean(),
diff --git a/convex/together.ts b/convex/together.ts
index 6c79ac8..a1c3f66 100644
--- a/convex/together.ts
+++ b/convex/together.ts
@@ -128,6 +128,14 @@ export const saveSummary = internalMutation({
     await ctx.db.patch(id, {
       generatingActionItems: false,
     });
+
+    // Generate TTS audio of the summary using CAMB AI
+    if (summary && summary !== 'Summary failed to generate') {
+      await ctx.scheduler.runAfter(0, internal.camb.generateSpeech, {
+        id,
+        text: summary,
+      });
+    }
   },
 });