Speech-to-text
The mechanics of turning audio into reliable transcripts.
Speech-to-textReal-time vs batch transcriptionStreaming and pre-recorded, and when to use each.Speech-to-textStreaming speech recognitionLive transcripts over a WebSocket.Speech-to-textPartial vs final resultsWhy live text rewrites itself.Speech-to-textWhat is voice activity detection (VAD)?Telling speech apart from silence and noise.Speech-to-textWhat is endpoint detection?Deciding when an utterance is over.Speech-to-textVAD vs endpointing vs turn detectionThree detectors that get confused for each other.Speech-to-textWhat is speaker diarization?Who spoke when.Speech-to-textDiarization vs identification vs verificationRelated speaker tasks with different answers.Speech-to-textSpoken language identificationDetecting which language is being spoken.Speech-to-textCode-switching in speech recognitionWhen speakers change language mid-sentence.Speech-to-textWord timestamps and forced alignmentPlacing every word on the timeline.Speech-to-textConfidence scores in speech recognitionWhat a 0.87 does and does not tell you.Speech-to-textPunctuation, casing, and inverse text normalizationFrom 'twenty euros' to €20.Speech-to-textHallucinations in speech recognitionWhen a model writes words nobody said.Speech-to-textSpeech-to-text latencyMeasuring the delay that users actually feel.Speech-to-textAudio formats for speech recognitionSample rates, codecs, and channels.Speech-to-textTranscribing noisy audioFar-field, overlap, and real-world rooms.Speech-to-textTelephony transcription8 kHz audio, codecs, and call recordings.Speech-to-textLong-form transcriptionHours of audio without drift or loss.Speech-to-textCaptions and subtitles: SRT, VTT, and timingTurning transcripts into readable captions.Speech-to-textIntegrating a speech-to-text APIKeys, sessions, WebSockets, and retries.
Try Reson8 on your own audio
Speech recognition for European languages that adapts to your vocabulary from text, in under a minute. Free tier, EU-only processing.