Show HN: We are 2.4x faster than Open AI latest SST model

Speech-to-Text technology is crucial for audio processing applications, but most solutions are limited by speed, accuracy, or feature constraints. What if you could transcribe audio with higher accuracy, better features, and significantly faster performance? We benchmarked OpenAI's newest gpt-4o-transcribe model against JigsawStack STT across real-world scenarios, including performance, accuracy, feature set, and multilingual capabilities. Here's what we found: Performance: - JigsawStack processes audio ~2.4x faster than OpenAI's model across all audio lengths - This speed advantage remains consistent with both short clips and longer audio files File Support: - OpenAI limits files to 25MB and 25 minutes of audio per request - JigsawStack handles files up to 100MB and 4 hours of audio per request Advanced Features: - Speaker Recognition: JigsawStack accurately identifies up to 50 different speakers with timestamps - Detailed Timestamps: Provides sentence-level timing data by default - Translation: Built-in support for translating audio into 100+ languages Real-world Accuracy: - Achieved 100% accuracy on noisy audio samples where OpenAI scored 94% - Properly identified repeated phrases and sentence boundaries that OpenAI missed Our team has focused on building a specialized solution that outperforms in nearly every metric that matters for real-world applications. Check out the full breakdown here: http://jigsawstack.com/blog/openai-audio-stt-vs-jigsawstack-stt http://jigsawstack.com/blog/openai-audio-stt-vs-jigsawstack-... Let us know what you think! We'd love feedback from the HN community - particularly from those working with audio transcription at scale.