SShortSingh.
Back to feed

VoiceLoop: Open-Source Browser Voice Agent Library Rivals OpenAI Realtime on Latency

0
·6 views

TodoForAI has open-sourced VoiceLoop, a zero-dependency JavaScript library that runs a full voice agent pipeline — VAD, speech-to-text, LLM, and text-to-speech — entirely in the browser. The library is designed for low latency, with first audio responses delivered in under a second by streaming TTS output while the LLM is still generating. A key feature is real barge-in detection based on transcribed words rather than microphone energy, which prevents the agent from cutting itself off due to speaker echo. Independent benchmarking via the team's own voice-agent-bench tool showed VoiceLoop with Deepgram and ElevenLabs matching OpenAI Realtime at roughly 862ms voice-to-voice latency while recording zero self-interruptions across 30 echo-coupled turns. The library is available on npm and supports any OpenAI-compatible LLM endpoint, with local WASM-based options requiring no API keys.

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)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Developer releases ChibiRigKit, an open-source tool to animate 2D characters from one image

A developer has released ChibiRigKit, a free open-source tool published on GitHub under the MIT License that converts a single character illustration into an animated 2D rig. The tool uses AI — specifically Codex CLI — during the creation pipeline to separate character parts, generate expression variants, and prepare animation assets, but the finished character runs entirely on standard browser technology without any ongoing AI dependency. Animations supported include blinking, eye and head movement, hair and clothing motion, and background replacement, with WebM export available. To address the challenge of realistic head rotation without full 3D modelling, ChibiRigKit interpolates between nine directional face configurations, each manually editable in the browser. The project is designed so that AI acts as a creation aid rather than a runtime requirement, allowing already-generated characters to continue working even if the underlying AI service changes or becomes unavailable.

0
ProgrammingDEV Community ·

Snowflake Iceberg Tables Let You Own Your Data Without Ditching Managed Services

Snowflake's managed Iceberg tables allow engineers to store actual Parquet data files in their own cloud storage (S3 or GCS) while still letting Snowflake handle metadata management, compaction, and schema evolution. Traditional Snowflake tables use a proprietary format that locks data behind the platform, forcing costly UNLOAD operations and creating catalog drift when teams try to integrate external tools like AWS Glue or Athena. By declaring an ICEBERG table with an external volume at creation time, teams decouple storage from compute, meaning other engines like Trino or DuckDB can query the same data directly without any export jobs. This approach reduces vendor lock-in risk, since migrating away no longer requires a complex data extraction strategy — the open Iceberg spec is already readable by standard query engines. However, experts caution that managed Iceberg tables involve trade-offs and are not suitable as a blanket replacement for every table in a Snowflake environment.

0
ProgrammingDEV Community ·

EU Child Safety Panel Delivers Report Pushing Age Verification and Stricter Platform Rules

The European Commission received the final report from its Special Panel on Child Safety Online on July 13, 2026, presented directly to President Ursula von der Leyen. The panel, established following von der Leyen's 2025 State of the Union address, drew expertise from health, neuroscience, child rights, and digital literacy, and also incorporated input from young people. The report signals a clear policy direction toward stricter platform safeguards, age verification schemes, and age-appropriate access controls for minors on social media. While the report is intended to inform future EU policy rather than impose immediate binding obligations, an EU Age Verification Scheme has been identified as a central area of consideration. Businesses operating digital services used by younger audiences are advised to begin reviewing their current practices ahead of formal legislative measures expected through 2026 and beyond.

0
ProgrammingDEV Community ·

ML Agent Reported Job Canceled While GPU Cluster Kept Running

An ML operations agent case study highlights a critical gap between agent-level cancellation and actual job termination on an external GPU scheduler. When an operator instructed the agent to cancel a fine-tuning run, the agent halted its own orchestration and reported success — but the job request had already been handed off to the scheduler. The scheduler subsequently accepted the job, which began consuming reserved compute despite the operator believing no allocation was active. The agent also risked losing the job ID needed to locate and properly cancel the workload. Experts recommend agents treat such cancellations as pending until the external scheduler confirms the job's final state, and actively retrieve the job ID to request and verify termination.