The problem with audio only
Subtitles are text that appears with a picture. A podcast or radio-style show has no picture. That means there is nothing to attach subtitles to, and a listener who does not speak the show’s language has no easy way in. The question for creators is how to give those listeners something they can follow.
The usual answer is to create a text or video form of the show first, and then translate that. Two routes are common: a transcript and a video version with subtitles. Each adds something different.
Route one: the transcript
A transcript is the written form of what was said. Once a transcript exists, it can be translated and published alongside the episode, on a web page or as a downloadable file.
What this adds:
- A reading option. Listeners who read the translation can follow at their own pace and reread passages.
- Searchability. Text can be searched, which helps people find episodes, quotes and topics.
- Access for deaf and hard of hearing people. A same-language transcript meets a need that translation alone does not.
- A base for other work. The cleaned transcript can later be used for subtitles, summaries or show notes.
What to be careful about: a transcript is not the same as subtitles. It has no timing and is often edited for reading. The differences between a transcript and subtitles matter when you decide which to produce. A transcript is better for reading; subtitles are better for following the audio as it plays.
Route two: a video version with subtitles
Many podcasters publish a video version, which might be a recording of the conversation, a still image or simple moving graphics with the audio. Once there is a video, translated subtitles can be added just as with any other video.
What this adds:
- Following along while listening. The viewer hears the original voices and reads their language, so they get the speakers’ tone and personality.
- Reach through video platforms. Where listeners search for video rather than audio, a subtitled version may be easier to find. Whether this brings more listeners is uncertain.
- Silent viewing. Many people watch videos with the sound off, so a short clip with translated text can introduce the show to people who would not otherwise try it.
- Language choice. Several subtitle tracks can be attached to one video, giving listeners their pick.
Subtitles for a video made from a podcast are bound by the same limits as other subtitles: lines must be short, text must stay on screen long enough to read, and speakers talking quickly may require condensing. Conversational shows are especially hard, because people interrupt each other, change direction mid-sentence and fill gaps with filler words.
How the translation is typically produced
- Start with a clean transcript. Podcast transcripts are often made with speech recognition, which is fast but can slip on names, technical words, accents and overlapping voices. A human pass improves them. Spoken-word translation tools may combine transcription and translation in one step, but errors in the first step carry over to the second.
- Decide what to translate. Spoken language is full of repetition and false starts. A translator may tidy these to make readable text, though the show’s style should still come through.
- Handle speakers. In a conversation, readers need to know who is talking. Speaker labels in a transcript and speaker identification in subtitles both help.
- Time the text. For subtitles, each line needs a start and end time matched to the speech.
- Check names and terms. Hosts, guests, places and specialist words should be spelled consistently. A short term glossary maintained for the show saves time on every episode.
- Review. A fluent reviewer should read the translation, particularly for shows with interviews or expert guests where a wrong word could misrepresent someone.
Machine translation can make this affordable for creators with small teams, but the output should be treated as a draft. Humor, sarcasm, slang and cultural references in conversation are frequent trouble spots.
What translation adds for listeners
- Listeners in other languages can follow shows on topics they care about.
- Language learners can use the original audio and the translation together, as long as they are aware that translations may be freer than the spoken words.
- A show with an international guest list can serve all those guests’ communities.
- Educational and informational shows become more useful to people who study or work in a second language.
None of this guarantees a larger audience. It removes one barrier, and the content must still be found, appreciated and shared.
Practical considerations
- Effort per episode. Each episode needs a transcript, translation, review and, for video, timing and export. A weekly show multiplies this, so many creators translate selected episodes or a back catalog of best-loved ones.
- Rights and permissions. Guests should know their words may be translated and published, and music or clips used in an episode may carry restrictions on where they can appear.
- Privacy. If recordings contain personal information, check how online tools handle uploaded audio.
- Sensitive subjects. Interviews about health, law or personal difficulties call for extra care and human review.
- Keep versions together. When an episode is corrected, update the transcript and subtitles in every language.
Summary
Audio-only shows cannot carry subtitles by themselves, but a transcript or a simple video version provides the base. Translating from there lets a show reach listeners who would otherwise be left out. The quality of the transcript, the care of the review and a realistic plan for the workload decide how well it works.