Move voice conversion to ffmpeg on hass

README.md

@@ -56,7 +56,6 @@ The bot is hosted on a VPS located at gateway.hackem.cc. The service, embassy-ap
 
 Node v20.10.0
 All main dependencies in the cloud and internal service are installed using npm i
-[Deprecated] To convert the doorcam stream to jpg, you need to install ffmpeg on the bot service and add it to PATH.
 
 ## Local deployment
 

api/embassy/routers/devices.ts

@@ -27,9 +27,9 @@ function linkMacPage(mac: string) {
             <a onclick="event.target.innerText = 'You can close this window now'; window.close();" \
             style="display:block;height: 100dvh;margin: 0;box-sizing: border-box;text-align: center;padding: 35vh 10vw;background: brown;text-decoration: none;font-size: 9vw;color: white;box-shadow: 0 0 10vw black inset;text-shadow: 0 0 1vw #000000;" \
             href=https://t.me/${embassyApiConfig.tgbot.username}?start=setmac__${mac.replaceAll(
-        ":",
-        "-"
-    )}>Press to set your mac to ${mac}</a>\
+                ":",
+                "-"
+            )}>Press to set your mac to ${mac}</a>\
         </body>\
     </html>`;
 }

config/default.json

@@ -151,7 +151,8 @@
             "entity": "media_player.lenovodash",
             "ttspath": "/api/services/script/announce_in_space",
             "playpath": "/api/services/media_player/play_media",
-            "stoppath": "/api/services/media_player/media_stop"
+            "stoppath": "/api/services/media_player/media_stop",
+            "voicepath": "/api/services/script/convert_voice_and_play"
         },
         "browser": {
             "target": "aee0b5cb4cd6bef1303d9e2c4be1a22e",

config/schema.d.ts

@@ -220,6 +220,7 @@ export interface SpeakerConfig {
     ttspath: string;
     playpath: string;
     stoppath: string;
+    voicepath: string;
 }
 
 export interface DevicesConfig {

deploy/Containerfile

@@ -17,8 +17,6 @@ RUN npm run build
 FROM docker.io/library/node:lts-alpine
 WORKDIR /app
 
-RUN apk add --update --no-cache ffmpeg
-
 COPY --from=build /app/node_modules ./node_modules
 COPY --from=build /app/dist ./
 

deploy/embassy-api/docker-compose.yml

@@ -10,6 +10,7 @@ services:
             - data-volume:/app/data/db
             - ./sec:/app/config/sec
             - ./static:/app/static
+            - /root/.ssh:/root/.ssh
         env_file:
             - .env
         restart: unless-stopped

services/hass.ts

@@ -1,15 +1,10 @@
-import path from "node:path";
-import { promises as fs } from "fs";
+import os from "os";
 
 import config from "config";
 import fetch, { Response } from "node-fetch";
 
 import { CamConfig, EmbassyApiConfig } from "@config";
-import { getBufferFromResponse } from "@utils/network";
-import { downloadTmpFile } from "@utils/filesystem";
-import { convertMedia } from "@utils/media";
-
-import { EmbassyBaseIP } from "./embassy";
+import { getBufferFromResponse, runSSHCommand } from "@utils/network";
 
 // Configs
 const embassyApiConfig = config.get<EmbassyApiConfig>("embassy-api");
@@ -90,33 +85,23 @@ export async function sayInSpace(text: string): Promise<void> {
     if (response.status !== 200) throw Error("Speaker request failed");
 }
 
-async function serveStaticFile(localPath: string, urlPath: string): Promise<string> {
-    const staticRootPath = path.join(__dirname, "..", embassyApiConfig.service.static);
-    const staticFilePath = path.join(staticRootPath, urlPath);
-
-    await fs.mkdir(path.parse(staticFilePath).dir, { recursive: true });
-    await fs.copyFile(localPath, staticFilePath);
-
-    return `${EmbassyBaseIP}/${urlPath}`;
-}
-
 export async function playInSpace(link: string): Promise<void> {
-    const requiresConversion = link.endsWith(".oga");
-
-    let linkToPlay = link;
-
-    if (requiresConversion) {
-        const { tmpPath, cleanup } = await downloadTmpFile(link, ".oga");
-        const convertedFilePath = await convertMedia(tmpPath, "mp3");
-
-        linkToPlay = await serveStaticFile(convertedFilePath, `/tmp/${Date.now()}.mp3`);
-
-        cleanup();
+    if (link.endsWith(".oga")) {
+        // ignore errors, ffmpeg writes to stderr
+        await runSSHCommand(
+            "hass.lan",
+            22269,
+            "hassio",
+            os.homedir() + "/.ssh/hass",
+            `wget -O /media/tmp/voice.oga ${link}`
+        ).catch(() => null);
+        await postToHass(embassyApiConfig.speaker.voicepath, {});
+        return;
     }
 
     const response = await postToHass(embassyApiConfig.speaker.playpath, {
         entity_id: embassyApiConfig.speaker.entity,
-        media_content_id: linkToPlay,
+        media_content_id: link,
         media_content_type: "music",
     });
 

utils/network.ts

@@ -104,6 +104,19 @@ export async function arp(ip: string, networkRange: string): Promise<string> {
     return mac;
 }
 
+export function runSSHCommand(host: string, port: number, username: string, key: string, command: string) {
+    const ssh = new NodeSSH();
+
+    return ssh
+        .connect({
+            host,
+            username,
+            port,
+            privateKeyPath: key,
+        })
+        .then(() => ssh.exec(command, [""]).finally(() => ssh.dispose()));
+}
+
 export class NeworkDevicesLocator {
     static async getDevicesFromKeenetic(routerip: string, username: string, password: string) {
         const ssh = new NodeSSH();