Saidwire
Live speech-to-text for events with verification within a second
Saidwire transcribes live events into verified text in real-time, showing words flipped to SAID about 0.1 seconds after audio is received.
About this launch
Saidwire provides real-time speech-to-text transcription for live events with multiple layers of verification. It covers events like rallies, press conferences and earnings calls, processing speech into verified words and quotes within about a second of being spoken. Multiple speech engines listen to the same audio, with words only counted as verified when two engines agree or a human confirmer validates it. For traders, words flip to SAID about 0.1 seconds after audio arrives. For newsrooms, it provides live transcripts, running stories and ready-to-post drafts while speakers are still talking. The service offers both real-time streaming via WebSocket and pre-recorded file transcription, with APIs compatible with OpenAI SDKs.
What it does
- Real-time streaming via WebSocket with interim and final results
- Pre-recorded file transcription supporting WAV, MP3, FLAC, Ogg/Opus, WebM, M4A/MP4
- OpenAI-compatible API endpoint for existing SDKs
- Punctuation and capitalization in transcripts
- Multiple response formats: JSON, text, verbose JSON, SRT, VTT
- Voice activity detection and speech endpoint detection
- Early word streaming for real-time keyword detection
- Multi-engine verification: words validated by two engines or human confirmer
Pricing
Pricing model: Paid
Event pass: $5 per event or $12 for 3 events. Saidwire Pro: $49/week for newsrooms and clip accounts (first event free).
Shopper questions
How fast is Saidwire's transcription?
Words are flagged a median of about 0.1 seconds after the audio reaches Saidwire, usually before the speaker has finished the word, with about one word in ten taking up to 0.8 seconds. The production engine returns a word about 25 ms after each 50 ms slice of audio arrives.
What audio formats does the API support?
The API supports WAV, MP3, FLAC, Ogg/Opus, WebM, M4A/MP4, and raw PCM for pre-recorded files. For streaming, formats can be detected automatically or specified via encoding parameter.
What file size and duration limits apply?
Pre-recorded file transcription has limits of 100 MB per request and 60 minutes of audio. URL downloads time out after 30 seconds.
Can I use Saidwire with existing OpenAI SDKs?
Yes, any SDK or tool that lets you set a custom base URL for speech-to-text works with Saidwire by changing the base URL to https://api.saidwire.com/v1 and providing a Saidwire API key.
How does Saidwire verify transcription accuracy?
Multiple speech engines listen to the same audio, and a word only counts as verified when two engines agree or a human confirmer validates it, with voice matching to ensure the main speaker is identified.
Saidwire has a free listing. The owner can unlock the full profile with a paid launch for $5:
- Founder story (locked)
- Milestones (locked)
- Demo video (locked)
- Dofollow website link (locked)
Shoppers also viewed
- v2md
Voice & transcription
- Shepit
Voice & transcription
- SpeechText.AI
Voice & transcription
- convert.express
Voice & transcription
