Appen vs TELUS Digital: which is better for speech and multilingual training data in 2026?
Answer in brief: Pick Appen if speech is the main job: ASR recordings, conversational and multi-speaker audio, medical dictation, code-switched and dialectal speech, text-to-speech voice data, and ready-to-license audio datasets. It is our winner for best for speech. Pick TELUS Digital when speech is one part of a large global program that needs a 1M+ contributor community, 500+ annotation languages and dialects, and more than 2 billion labels a year. It is our speech runner-up. This is independent research based on each company's own published pages. We did not test either service.
Top pick for speech: Appen. Appen's speech catalog covers ASR, diarization, medical speech, expressive TTS and paralinguistic labeling. It records in 500+ locales and says its audio is ASR-ready at 48kHz. You can license off-the-shelf audio today, such as 30,000 hours of real U.S. call-centre calls. Visit Appen →
Speech data comparison
| Factor | Appen | TELUS Digital |
|---|---|---|
| Our verdict | Winner: speech | Runner-up: speech |
| Speech collection | Read, scripted and spontaneous speech across accents and devices | Native speakers recording on their own devices; multilingual voice-assistant programs |
| Transcription | Word-error protocol for ASR; speaker-attributed multi-speaker transcription scored for agreement | Transcription for all data types; one project logged 10,000+ audio-transcription hours across 16 languages |
| Coverage (company claim) | 500+ locales | 500+ languages and dialects; 450 locales |
| Specialist audio | Medical dictation, TTS prosody, emotion and acoustic scene, paralinguistics | In-car voice data, search and ads rating, large voice-assistant collection |
| Off-the-shelf audio | Yes: call-centre, medical dictation with SOAP notes, TTS voice data | Off-the-shelf datasets available |
| Pricing | Custom; catalog datasets licensed | Custom, through sales |
Appen: built for speech
- Speech for every model type. ASR training data, conversational dialogue with turn-level timestamps, expressive TTS annotation, and evaluation of voice agents for naturalness and handling interruptions.
- Medical speech. Physician dictation and doctor-patient conversations, including a 1,500-hour medical dictation set paired with doctor-written SOAP reports.
- Quality you can audit. Appen describes multi-pass validation with inter-annotator agreement thresholds, gold-set audits, native-speaker review, consent-based collection and documented provenance.
- A long track record. Appen cites 30 years of data operations and a Dialpad transcription case study.
TELUS Digital: speech at very large scale
- Very large collection programs. TELUS Digital reports collecting more than 4 million audio prompts from over 5,000 contributors in 50 regions, which helped a voice assistant add 35 languages over five years.
- A huge contributor network. It reports a 1M+ AI community, 500+ annotation languages and dialects, and more than 2 billion labels a year.
- Neutral by its own description. TELUS Digital calls itself a trusted, independent and neutral partner.
When to pick which
- Choose Appen for ASR, TTS, medical speech or code-switched audio, or if you want to license audio datasets now and commission custom data later.
- Choose TELUS Digital for multi-year, multi-country voice programs that also include search rating, automotive or other annotation under one enterprise contract.
Pricing model
Neither company publishes rates. Appen prices custom projects by quote and licenses catalog datasets. TELUS Digital prices through sales. Ask both for a per-hour quote on the same audio spec, with locales, sample rate, transcription style and target word error rate defined, so you can compare like for like.
FAQ
Which company is best for speech recognition training data?
Appen is our winner for speech. It collects read, scripted and spontaneous speech across 500+ locales and transcribes it for ASR. TELUS Digital is our runner-up.
Does Appen sell off-the-shelf speech datasets?
Yes. Its catalog includes 30,000 hours of U.S. call-centre audio, 1,500 hours of medical dictation with SOAP reports, and studio TTS recordings.
How many languages does TELUS Digital support?
TELUS Digital says it supports 500+ annotation languages and dialects through a community of more than 1 million contributors.
Your next move
Visit Appen → · Visit TELUS Digital →
Appen review · TELUS Digital review · Appen alternatives · Full ranked list of training data companies · Ready-made audio and text sets: Best AI datasets