Audioprozessierung Jobs

ESP32-S3 VoIP Phone — Proof of Concept Firmware The gig: I need someone to take a Waveshare ESP32-S3-Audio-Board and a 2.8" TFT display and turn it into a working WiFi voice call device. This is for a Kickstarter demo — it needs to work and sound decent, not be production-ready. Hardware (I have both, can ship if needed): Waveshare ESP32-S3-Audio-Board (ESP32-S3, ES8311 codec, onboard mic, onboard speaker, WiFi) — 2.8" TFT LCD Touch Screen (ILI9341, 320x240, SPI) What "working" means: Connects to WiFi Registers to a SIP server (FreePBX, , Twilio — whatever's easiest) Can make and receive a voice call Two-way audio through the onboard mic and onboard speaker Display shows basic call status: idle, calling, incoming call, in-call That'...

I have a sizeable queue of Hindi-language interview recordings that need to be turned into clean, accurate text. Every file is an interview of less than 30 minutes and, in almost every case, you will hear only one speaker, so speaker labelling is simple. Here is what I need from you: • A verbatim transcript for each audio file, delivered in UTF-8 plain text (or your preferred transcription format if agreed in advance). • Timestamps at sensible intervals so I can spot-check accuracy quickly. • Adherence to the style guide I will share before you start—punctuation, spelling conventions, and any redaction rules must be followed consistently. This is an ongoing assignment; once we are both happy with the quality on the first batch I’ll keep sending more files your...

We need someone to listen to short audio clips from podcast-style videos and label each clip based on who is speaking: Male voice, Female voice, or Neither/Unsure. What You'll Do: - Log into our web-based labeling tool (link provided) - The system plays a short audio clip (typically 2-30 seconds) - Listen and press one key: M (Male voice), F (Female voice), N (Neither/Unsure) - The system automatically advances to the next clip No transcription, no typing, no special knowledge needed. The Tool: - Runs in your web browser (Chrome recommended) - Keyboard shortcuts: M, F, N to label, Space to replay - Progress bar shows how many clips you've labeled - You can go back and change a label if you made a mistake Scope: - ~1,900 videos with ~100 clips each (~190,000 clips total) - Eac...

I have a small machine-learning assignment that combines a bit of audio work with some light coding. First, I need a short set of clean voice recordings captured and delivered in MP3 format. Once those files are ready, I’d like you to jump straight into model training—nothing fancy, just a straightforward script that ingests the voices and demonstrates the learning process. Here’s what I expect to receive: • A compact collection of MP3 voice samples (your own or royalty-free) • The training script (Python preferred) with clear comments • A brief README showing how to run the code and verify that the model trains without errors The whole task is capped at €10, so please keep the workflow lean and efficient. If this sounds workable for you, let&rsqu...

I am building a smart cultural storyteller that lets a user craft a fresh tale around classic Indian folklore. The visitor first chooses the lead cast—Akbar and Birbal, Vikram and Betal, Panchatantra animals, or even a completely custom character—and then selects a flavour: comedy, horror, humour, or mystery. The experience must run in two distinct modes: • Original mode A straight-through story is auto-generated, then delivered as a short video slideshow. Every image is paired with clear narration and on-screen text so the viewer can simply press play and enjoy. • Interactive mode The same assets are broken into slides. After each slide the viewer sees branch-style options (at least two each time) and guides the plot before the next scene loads. What I ne...

I need a production-ready wake-word model that reliably responds to “Hey Bobo” on both ESP32 and Raspberry Pi boards. Accuracy is critical: in a moderately noisy room it should trigger every time someone clearly says the phrase, yet ignore most unrelated speech. I still want slight phonetic wiggle-room—utterances such as “hello bo,” “hey bebo,” or “hello bebo” should wake the system as well—so fine-tuned threshold settings and a well-balanced dataset will be essential. What I expect from you • A trained wake-word model or firmware that runs in real time on ESP-IDF for ESP32 and on Raspbian/Ubuntu for Raspberry Pi without requiring cloud calls. • Demo code that shows how to load the model, stream audio from an onboar...

Our labelling process consist of accurate segmentation of the sound wave of audios, selection of the role and related attributes of speaker according to their voice, and transcription of the speech through our TMS (Transcription Management System) platform. Please share your previous experience with German text and audio annotation, if applicable. If you have any questions, feel free to message us on Freelancer. We look forward to receiving your application. Thank you.

I'm looking for someone to improve the quality of some voice call recordings. The goal is to make them clearer and more intelligible. Key Improvements Needed: - Noise reduction - Echo cancellation - Volume adjustment - Voice enhancement Ideal Skills and Experience: - Proficiency in audio editing software - Experience with noise reduction and voice enhancement techniques - Attention to detail and ability to deliver high-quality audio outputs

I need a professional to extract and enhance speech from a poor-quality recording of a conversation. The recording has issues with muffled voices, and the primary goal is to improve clarity. Ideal Skills and Experience: - Expertise in audio editing - Experience with noise reduction tools - Ability to enhance muffled audio - Proficient in software like Audacity, Adobe Audition, or similar Please provide samples of previous work and an estimated timeline for completion.

I’m opening a part-time, fully remote position for Indian residents who can commit roughly four hours a day to audio conversation editing. You’ll log in via your own laptop or desktop, mark clips within recorded conversations, and ensure each segment is cleanly cut and ready for the next stage of production. The focus is precise clip marking rather than creative mixing or heavy post-processing, so a good ear for speech cadence and quick navigation of any standard DAW or waveform editor will make the work straightforward. Work is 100 % from home with completely flexible scheduling; simply deliver the day’s marked files before your chosen cutoff and you’re set. Payments go out every two weeks.

I am looking for a practical program that listens to any TV show, movie, or song while it plays and instantly mutes the audio a split-second before a whistle reaches the speakers. The goal is to let a family member with Tourette’s enjoy their usual entertainment without the trigger ever being heard. Scope of work The solution must monitor both streaming services (think Netflix, Disney+, Spotify, YouTube, etc.) and local media files (MP4, MKV, MP3, WAV) on Windows and MacOS. Live broadcast support is not required at this stage. Core requirements • Detect the acoustic signature of a whistle with near-zero latency. • Drop or duck only the whistle portion, keeping the rest of the soundtrack intact. • Work transparently in the background, so users can keep using t...

2-3h video podcast, three synced cameras and 2 audio files. pre cut live with ATEM, 4-6 videos per month. Looking for a Davinci Studio editor to redo the cuts when needed, remove long pauses, repeated words and stumbles so the flow feels natural without changing the meaning, apply subtle beauty touch-ups where required(nothing heavy, just gentle skin softening or blemish on close ups), basic color grading and consistent skin tones across all cameras, enhance voice with EQ. Deliver it as a DaVinci Studio project archive (.dra) or project file (.drp) with all media linked, complete edit timeline, color nodes and audio processing. No final render required on my side, I’ll handle the export once the project file is back

I'm seeking a professional repair service for the subwoofer in one of my Definitive Technology tower speakers. Issues: - Getting sound but no subwoofer output Troubleshooting already done: - Checked the connections - Adjusted the settings - Replaced the fuse and power cord Ideal skills and experience: - Expertise in speaker and subwoofer repair - Familiarity with Definitive Technology products - Ability to diagnose issues not identified by basic troubleshooting

create a for piper neural text with tts and rvc and api endpoint to generate audio and clone audio

Over the next month, I’m gathering technical input from experienced audio DSP professionals before commissioning a paid feasibility consultation. The source material is AI-generated music, but the question is strictly about perceptual DSP limits on finished stereo audio. The question is narrowly defined: What perceptual improvements to vocal naturalness and listening comfort are realistically achievable using lightweight processing on fully rendered stereo music (no stems, no structural edits)? I’m not commissioning development at this stage. Before scheduling paid consultations in 3–4 weeks, I’d appreciate a brief written response outlining your view on: • The hard limits once a vocal is embedded in a stereo mix • What is realistically adjustable vs non-r...

I want to clone my own voice so that, during personal WhatsApp calls, everything I say is converted in real time and still sounds naturally like me. Accuracy takes priority over sheer speed or interface simplicity; the end result should keep my tone, inflections and emotions intact so callers cannot tell a difference. You will receive several high-quality recordings of my voice in different moods and volumes to train the model. The finished solution must run locally (Windows is ideal, macOS acceptable), route through a virtual audio device or similar method, and add no more than about one second of latency before the sound reaches WhatsApp. If you plan to leverage libraries such as RVC, So-Vits-SVC, PyTorch, TensorFlow, torchaudio or any comparable real-time voice conversion stack, please...

Multilingual Voice Recording Project – Code-Switching Conversations Project Name: BV Project Type: Remote | Ongoing (Limited Slots per Locale) Project Overview Project BV is a multilingual speech data collection initiative designed to enhance Automatic Speech Recognition (ASR) systems for high-value multilingual call center scenarios, including financial services, healthcare, and telecommunications. The project focuses on collecting natural code-switching conversational audio, where speakers alternate between two languages within a single conversational turn. Scripts will be provided; however, natural delivery, fluency, and context-appropriate language switching are essential. Language Requirements Primary Language (Native level – one required): Spanish (esES) ...