Speech to text: AI tools compared (15)

What speech to text tools are for

Speech-to-text tools convert spoken material into writing from a file, microphone, or live conversation. They can support journalists, researchers, teams, and people who prefer dictation to typing. Depending on the service, you upload a recording, separate speaker contributions, or dictate into an application. Previously, finding a decision in an interview meant listening through the recording and copying the relevant passage. With a transcript, you can search for a term, locate the sentence, and return to the audio to check it. AssemblyAI supports file and live transcription with speaker distinction. Vibe can transcribe files on your computer. The practical change is how you access and navigate spoken content. You still need to verify important wording, confirm who said what, and protect both the source recording and the written material when the conversation contains confidential information.

Example uses:

• Transcribe a recorded interview to locate passages through text search, then verify the wording against the original audio.
• Dictate a draft into an application when typing would be inconvenient for the task you are working on.
• Turn a recorded lesson into searchable written material, then correct personal names and terminology specific to the subject.
• Separate speaker contributions in a conversation to help review the decisions and disagreements expressed during the discussion.
• Transcribe a file locally with a suitable tool when your organization requires recordings to remain on your computer.

Our picks by need

These picks follow simple rules applied to the verified data, with no partnership or affiliate link:

• Best scored : Whisper. 95/100 on our access and transparency grid.
• Cheapest paid option : AI Dictation. First plan at $84.99 per year.
• To keep data in Europe : Vibe. Processing on your device.

The main tools at a glance

Whisper 95/100

Whisper transcribes audio recordings, recognizes the spoken language and translates recordings into English. It serves people who want to obtain transcriptions or produce subtitles with timestamps from audio. Features: Transcribe an audio recording; Recognize the spoken language; Translate a recording into English. Entry price: $0.01 pay as you go, with a free tier. Runs on: Windows, macOS, Linux, and API. Data hosting: Several regions.

AssemblyAI 78/100

AssemblyAI is for people transcribing audio files, videos, or live conversations. It supports speaker identification, timestamped subtitles, transcript summaries and analysis, and masking personal information in content. Features: Transcribe audio and video files; Transcribe live conversations; Distinguish between speakers. Entry price: $0.15 pay as you go, with a free tier. Runs on: web browser and API. Data hosting: Several regions.

Rev 75/100

Rev lets people working with audio and video recordings transcribe them and create subtitles. They can request human verification, correct text and speaker names, and extract and share recording excerpts. Features: Transcribe audio and video recordings; Have a transcription checked by a person; Create subtitles. Entry price: $0.25 pay as you go, with a free tier. Runs on: web browser, iPhone and iPad, Android, and API. Data hosting: United States.

Wispr Flow 73/100

Wispr Flow serves people dictating text into applications. It removes hesitations, adds punctuation, remembers personal terms, inserts saved text through voice shortcuts and transcribes and summarizes meetings. Features: Dictate text into applications; Remove hesitations and add punctuation; Remember personal names and terms. Entry price: $12 per month, with a free tier. Runs on: Windows, macOS, iPhone and iPad, and Android. Data hosting: United States.

Vibe 68/100

Vibe serves people transcribing audio and video on their computers. It processes multiple files, transcribes microphone or system audio, creates subtitles, summarizes transcripts and translates speech into English. Features: Transcribe audio and video files on your computer; Process multiple files in sequence; Transcribe microphone or system audio. Free tier available. Runs on: Windows, macOS, Linux, and API. Data hosting: On device.

Deepgram 65/100

Deepgram helps people working with speech transcribe audio files and live speech. It distinguishes speakers, generates voices from text and supports the construction of voice agents through its platform. Features: Transcribe audio files; Transcribe live speech; Distinguish speakers. Entry price: $0 pay as you go, with a free tier. Runs on: web browser and API. Data hosting: Several regions.

HappyScribe 65/100

HappyScribe helps people working with audio recordings produce transcripts, subtitles, and translations. It also supports meeting summaries and lets users have a transcript reviewed by a professional. Features: Turn audio recordings into text; Create and correct subtitles; Translate transcripts. Entry price: £102 per year, with a free tier. Runs on: web browser, iPhone and iPad, Android, and API. Data hosting: European Union.

Sonix 60/100

Sonix lets people working with audio and video files transcribe them and correct text in a synchronized editor. It also supports transcript translation, subtitle export, and shared review by teams. Features: Transcribe audio and video files; Correct text in a synchronized editor; Translate transcripts. Entry price: $10 pay as you go. Runs on: web browser and API. Data hosting: United States.

Rylo 58/100

Rylo helps people who want to read call conversations display speech as text and have typed replies read aloud. It also transcribes in-person conversations and stores transcripts on the device. Features: Display speech from a call as text; Read a typed reply aloud; Transcribe an in-person conversation. Free tier available. Runs on: iPhone and iPad, Android, and macOS. Data hosting: Several regions.

WhisperAPI 58/100

WhisperAPI serves people transcribing audio or video files, including files supplied through URLs. It produces timestamped subtitles, automatically detects the language and offers a choice of transcription model size. Features: Transcribe an audio or video file; Process a file supplied through a URL; Produce timestamped subtitles. Entry price: $10 pay as you go, with a free tier. Runs on: web browser and API. Data hosting: Several regions.

The category in figures

Of the 15 tools in the category, 12 (80%) offer a free tier, 80% have a public API and 1 have a French interface. The median entry price, per month, is $9 (across 3 tools with a published price). 12 disclose where data is hosted, 4 of them in Europe or on device.

Also useful

These tools belong to another category but also cover this need:

• ElevenLabs ElevenLabs turns text into speech, transcribes recordings and produces sound effects. It serves people who wan
• Otter.ai Otter.ai transcribes meetings live, distinguishes speakers and summarizes discussions. It serves people who wa
• Read AI Read AI transcribes meetings, summarizes discussions and decisions and extracts action items. It serves users
• tl;dv tl;dv lets people attending online meetings record and transcribe discussions and produce meeting reports with
• Avoma Avoma helps teams record and transcribe meetings and calls while identifying participants. It summarizes decis
• Riverside Riverside records remote audio and video interviews for people producing podcasts or videos. It transcribes re
• Groq Groq provides API access to language models and transcribes audio recordings. It serves people who want to pro
• Fathom Fathom helps meeting participants record and transcribe discussions. It prepares summaries and action items, s
• Noota Noota helps people conducting meetings, calls, or interviews record and transcribe conversations. It also supp
• Descript Descript edits videos through changes to their transcripts and cleans up voice audio. It serves people who wan
• Tactiq Tactiq lets people attending Google Meet, Zoom, and Teams meetings transcribe and summarize discussions. It ex
• Cleanvoice AI Cleanvoice AI serves users processing audio recordings or podcasts. It reduces background noise, removes hesit

Choosing well and avoiding pitfalls

Test recordings that reflect your accents, microphones, and specialist vocabulary. Assess French recognition, names, punctuation, and passages with overlapping speech. Check speaker distinction, timestamps, and export formats against your intended use. For dictation, test compatibility with the applications where you write. For sensitive material, establish whether processing is local or remote and review file retention. Compare human review options and integrations where relevant. Finally, examine how recorded duration and live processing are billed. Include a difficult excerpt in your evaluation so the decision reflects your actual recordings rather than a clean demonstration.

A transcript can miss a negative, confuse a name, or assign a sentence to the wrong speaker. Noise and overlapping voices make checking harder. Listen again to important passages instead of treating fluent writing as proof of accuracy. Some systems may also produce text from silence or ambiguous sounds. Make sure the people concerned are appropriately informed about recording in your context, and restrict access to both the original audio and the resulting written files.

To compare with the cost of using an AI model directly, see the API price calculator.

Free or paid?

Free plans often let you test a recording or process a limited amount of material. Local software may also be available without a subscription, although hardware and setup requirements still matter. Paying becomes useful for regular processing, live transcription, integrations, or human review. Check storage terms and included export options alongside the transcription features themselves.

Tools in the category with a free tier: Whisper, AssemblyAI, Rev, Wispr Flow, Vibe, Deepgram, HappyScribe, and Rylo.

For which jobs?

The profiles that use these tools most, based on the use cases documented by publishers: podcasters, video creators, researchers, developers, lawyers, and students.