Building a Node.js Voice Assistant with ElevenLabs Text-to-Speech

·3 views

A DEV Community tutorial walks developers through creating a command-line voice assistant using Node.js and the ElevenLabs API. The assistant accepts typed queries, forwards them to an LLM such as OpenAI's ChatGPT for a reply, converts that reply into speech, and plays the audio back. The guide explains the prerequisites, including Node.js 18 or later, API keys, and optionally ffmpeg for audio playback. It also covers project setup, storing credentials in a .env file, and a lightweight text-to-speech wrapper that saves the output as an MP3. The article notes that ElevenLabs also supports voice cloning using a short audio sample.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

macOS Git Trusts Compromised eMudhra Certificates, Here's How to Remove Them

Apple's bundled git on macOS can still trust compromised eMudhra certificates, which may let automated builds accept revoked identities. The guide explains that macOS git reads certificates from the system keychain rather than OpenSSL defaults, so revoked certificates remain valid if they stay in the keychain. It provides commands to detect eMudhra certificates, delete them by SHA-256 fingerprint, and confirm the trust chain is clean. It also recommends adding a CI check that fails builds if the certificates reappear, and notes checking user keychains, Homebrew git configuration, and cached TLS sessions.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Build an n8n Watchdog That Alerts When a Workflow Never Runs

n8n's Error Trigger only catches failed runs, so a workflow that never starts, due to a dead cron, changed webhook URL, or deactivation after an edit, goes unnoticed. The article proposes a dead-man's switch: a separate watchdog workflow that runs hourly and checks the monitored workflow's execution history via the n8n Public API. It classifies the workflow as overdue if no runs exist or the latest run is older than a configurable threshold, and sends a Gmail alert only in that case. Setup requires an n8n API key, the target workflow ID, a threshold set to roughly twice its expected interval, and Gmail OAuth2 credentials. The author reports testing the detection logic on a local n8n instance but says email delivery still needs to be confirmed by running a test alert.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

FiftyOne 1.19 Adds Native MCAP Playback for Synced Drone Sensor Recordings

FiftyOne 1.19 can now open MCAP log files directly, combining camera, LiDAR, GPS, IMU, and log streams on one shared timeline without requiring a ROS installation. The author demonstrates this with a UAV flight through a Virginia forest lasting about four and a half minutes, which produced nearly 600,000 messages across 31 channels. Because MCAP is a self-describing container, each recording stays decodable even after the software that wrote it is gone. Users point a sample's filepath at a .mcap file, and the App automatically treats it as multimodal data, rendering image, 3D, map, plot, log, and message tiles that all move together during playback. Channels without a built-in decoder still appear as raw, collapsible record trees, so no data in the file goes unseen.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

MiniMax Raises About $2B and Releases M3 Model Usable Inside FiftyOne

Shanghai-based AI lab MiniMax, listed in Hong Kong since January, has announced a private placement and convertible-bond raise of roughly HK$16 billion, its first major fundraising since the IPO, with most funds directed to model research and infrastructure. Alongside this, the company released MiniMax-M3, a natively multimodal model with about 428 billion total parameters and a 1 million-token context window. A new FiftyOne plugin lets users apply the model to images and videos for detection, keypoints, captioning, visual question answering and other tasks without fine-tuning or hosting a model server. The plugin writes outputs as native FiftyOne label objects and records provenance such as prompts and sampling settings. The model uses a sparse attention mechanism that its developers report substantially cuts compute and speeds processing at long context lengths.

0 comments Read more at DEV Community